Published 2 July 2026 — Anthropic, anthropic.com

Anthropic Publishes Fable 5 Cybersecurity Safeguards and Proposes Industry Jailbreak Severity Standard

Anthropic has detailed the cybersecurity safeguards built into Claude Fable 5 and proposed an industry-wide framework for scoring AI jailbreak severity. The safeguards classify requests into four categories — prohibited (ransomware, malware, data exfiltration), high-risk dual use (penetration testing, exploit development), low-risk dual use (vulnerability identification, OSINT), and benign (secure coding, incident response). The proposed Cyber Jailbreak Severity (CJS) framework rates jailbreaks on four axes — capability gain, breadth, ease of weaponisation, and discoverability — producing scores from CJS-0 (Informational) to CJS-4 (Critical). Amazon, Microsoft, Google, and other partners are collaborating on the framework.

For Australian businesses using Claude in legal, compliance, or security roles, the published safeguards clarify what tasks are permitted and what is blocked, reducing ambiguity in AI governance and procurement decisions. If adopted industry-wide, the CJS scoring system would give security teams a standardised benchmark for assessing AI model risk.

Source: Anthropic, 2 July 2026 →

Published 30 June 2026 — Anthropic, anthropic.com

Anthropic Launches Claude Science, an AI Research Workbench With Grants Up to USD $30,000 for Scientific Projects

Anthropic has launched Claude Science, an AI workbench designed for scientific research, now in beta for Claude Pro, Max, Team, and Enterprise users on macOS and Linux. The platform pre-configures access to more than 60 specialised scientific databases and tools — including UniProt, PDB, Ensembl, ChEMBL, and GEO — and allows researchers to coordinate multi-step analyses, manage compute resources across laptops, HPC clusters, or on-demand GPUs, and produce fully reproducible artifacts with complete code history. A built-in reviewer agent checks citations and calculations within the workflow before finalising outputs.

Anthropic is offering up to 50 AI for Science grants, each providing up to USD $30,000 in Claude credits plus up to USD $2,000 in compute credits from Modal. The application deadline is 15 July 2026, with funded projects running from 1 September to 1 December 2026. Australian researchers at universities, hospitals, and non-profit scientific organisations are eligible to apply through the standard grant program; academic labs can also access a discounted Team plan. For Australian organisations in health, pharma, environmental science, or any research-intensive field, Claude Science represents a concrete, funded entry point to AI-assisted scientific workflows — with a grant deadline less than two weeks away.

Source: Anthropic, 30 June 2026 →

Published 30 June 2026 — Anthropic, anthropic.com

Anthropic Launches Claude Sonnet 5, Its Most Agentic Mid-Tier Model Yet

Anthropic has released Claude Sonnet 5 across all Claude plans and the Claude API, describing it as the most agentic Sonnet model yet, with substantial gains over Sonnet 4.6 in reasoning, tool use, coding and autonomous task completion while approaching Opus 4.8-level performance at a lower cost. Introductory pricing is US$2 per million input tokens and US$10 per million output tokens through 31 August 2026, rising to US$3 and US$15 after that date.

The model is available through Microsoft Foundry (Azure), AWS and Google Vertex AI, as well as via Claude Code and Cowork. Australian businesses already using Claude models inside Microsoft 365 Copilot or via cloud platforms can access the upgrade without new procurement, making it relevant to organisations evaluating AI tooling for coding, automation and knowledge-work tasks.

Source: Anthropic, 30 June 2026 →

Published 29 June 2026 — IAPP, iapp.org

Five Eyes Agencies Warn Frontier AI Will Accelerate Cyber Threats — Timeline Is Months, Not Years

The cybersecurity agencies of Australia, Canada, New Zealand, the United Kingdom, and the United States have released a joint statement warning that frontier AI models will “accelerate the speed, scale, and sophistication of cyber threats.” Citing urgency, the agencies stated: “The timeline is not years, it is months.” The statement, published on cyber.gov.au by Australia’s ASD alongside its Four Eyes counterparts, calls on business leaders to understand and assess risk and accountability, prioritise foundational cybersecurity practices and controls, empower cyber leaders with sufficient authority and resources, and stay actively engaged as both threats and guidance evolve.

For Australian businesses, the joint advisory from the ASD represents a direct government call to action, advising organisations to treat cybersecurity legal standards as an evolving baseline, accelerate retirement of legacy systems, reassess third-party vendor access controls, update incident response planning for higher-frequency AI-enabled attacks, and strengthen board-level governance. The statement concludes: “Success will come from getting the basics right, acting quickly, and integrating cyber security into core business strategy.”

Source: IAPP, 29 June 2026 →

Published 24 June 2026 — Microsoft Newsroom, news.microsoft.com

Microsoft AI in Education Report: 92% of Students Using AI But Most Without Formal Training — Workforce Implications for Australian Businesses

Microsoft’s third annual AI in Education Report, surveying 3,345 students and educators across six countries including Australia, finds that 92% have already used AI for educational purposes and 78% report increased use over the past year — yet 77% of students and 53% of educators have not received any formal AI training. 58% of education leaders report their institutions are actively implementing or scaling AI initiatives, while academic integrity concerns are cited by 41% of students and 42% of educators as their primary barrier. In response, Microsoft is rolling out new AI tools: Copilot Notebooks (AI-powered study workspaces for uploading and querying course materials), a Learning Agent now generally available to Microsoft 365 Copilot licence holders that delivers personalised AI upskilling for employees, and a Study and Learn Agent within Copilot Chat for research-based guidance.

For Australian businesses, the findings carry direct workforce implications. Australia’s new workforce entrants are heavy AI users but have largely developed AI habits informally and without enterprise-grade guardrails — meaning practices formed during education may not align with an organisation’s governance or data handling policies. The 53% of educators without formal AI training signals a compounding pipeline problem: the professionals shaping the AI habits of Australia’s next graduates are themselves under-equipped. Organisations that assume new recruits arrive with structured AI literacy are likely to be disappointed and may need to treat AI fundamentals as a standard part of induction alongside data governance and security awareness training.

Source: Microsoft Newsroom, 24 June 2026 →

Published 23 June 2026 — Anthropic, anthropic.com

Anthropic Launches Claude Tag for Slack: Async AI Agent for Enterprise Teams

Anthropic has launched Claude Tag, a Slack integration available in beta for Claude Enterprise and Team customers that places Claude as a shared AI member inside channels. Team members tag @Claude to delegate tasks; a single Claude instance serves all participants in a channel, builds context from channel history and connected tools, and works asynchronously across hours or days. Anthropic reports that 65% of its own product team’s code is now written by an internal version of Claude Tag, and use has expanded to sales analytics, support ticket management, and debugging.

Enterprise governance controls include channel-by-channel tool and data access restrictions, token spend limits at the organisation and channel level, and full activity logs with requestor attribution. The existing Claude in Slack app is being retired, with a 30-day migration window for current users. Australian businesses using Slack with Claude Enterprise or Team subscriptions can begin beta access immediately; organisations should verify data residency requirements and review channel-scoped permissions before enabling access to sensitive data sources.

Source: Anthropic, 23 June 2026 →

Published 17 June 2026 — Clayton Utz, claytonutz.com

Clayton Utz Warns Australian Boards of AI and Digital Supply Chain Blind Spots

Clayton Utz has released a white paper, “Closing the digital blind spot: managing third-party risk,” warning that 68% of global organisations reported increased digital supply chain disruptions in 2026, and that Australian boards remain largely unaware of vulnerabilities beyond their headline technology suppliers. The paper highlights that the 2024 CrowdStrike outage cost Australian businesses an estimated $1 billion, and that fourth and fifth-party provider failures now pose risks to SOCI Act-regulated critical infrastructure entities.

The report warns that organisations are shedding human expertise in favour of AI systems, creating key-person risk if those systems fail, while only 64% have formal processes to assess AI tool security before deployment. ASIC and APRA are actively targeting operational failures linked to agentic AI. Clayton Utz has published an 8-part Digital Resilience Toolkit to help Australian organisations move beyond compliance toward genuine resilience.

Source: Clayton Utz, 17 June 2026 →

Published 16 June 2026 — Microsoft Newsroom, news.microsoft.com

Aussie Workers Race Ahead With AI While Organisational Leadership Falls Behind

Microsoft Australia’s 2026 Work Trend Index, drawing on surveys of 2,000 Australian workers and trillions of anonymised Microsoft 365 productivity signals, finds 63% of Australian AI users are now producing work they could not complete a year ago — rising to 84% among the most advanced users. Yet only 28% report their organisation has a clear AI strategy, and just 13% are recognised for innovation when results take time. The research found organisational factors — culture, management and talent practices — drive twice the AI impact of individual effort (67% versus 32%).

Microsoft Australia President Jane Livesey noted the barrier is “no longer the technology — it’s whether leaders provide the clarity, culture, and confidence.” For Australian businesses, the implication is direct: deploying AI tools is not enough. Organisations must treat AI adoption as an education and enablement program, build reward structures that support experimentation, and close the widening gap between worker capability and leadership readiness.

Source: Microsoft Newsroom, 16 June 2026 →

Published 16 June 2026 — Microsoft Newsroom, news.microsoft.com

Sydney-Built AI Tutoring Platform Cogniti Scales Globally via Microsoft Marketplace

Cogniti, an AI platform built at the University of Sydney by Professor Danny Liu that lets educators create subject-specific AI tutoring agents, has been listed on Microsoft Marketplace since January 2026 and is now scaling internationally on Microsoft Azure. Leiden University has deployed it across four faculties to around 2,000 students, and nursing students at two New Zealand institutions logged 820 simulation conversations in one semester, with 92% reporting it improved their studies.

The platform’s path to market — an Australian university research project commercialised through a major cloud marketplace — is a template Australian software and edtech businesses can point to when pursuing global distribution for locally-built AI tools, and reflects Australia’s growing profile as an AI exporter.

Source: Microsoft Newsroom, 16 June 2026 →

Published 12 June 2026 — Anthropic, anthropic.com

US Government Orders Global Suspension of Anthropic’s Fable 5 and Mythos 5 Access

Anthropic received a US government export control directive on 12 June 2026 requiring the immediate suspension of all access to its Fable 5 and Mythos 5 models for every user globally, including employees. The government cited national security authorities and alleged awareness of a jailbreaking method that could bypass Fable 5’s safety guardrails — specifically, a technique involving asking the model to read a codebase and fix software flaws. All Fable 5 and Mythos 5 access was terminated immediately for customers worldwide. Other Anthropic models, including Claude Opus 4.8 and Haiku 4.5, remain unaffected.

Anthropic has publicly stated it disagrees with the decision, characterising the identified vulnerability as “a narrow potential jailbreak” and noting that the same capability exists in competing models. The company warned that applying this standard industry-wide “would essentially halt all new model deployments for all frontier model providers.” Australian businesses that have integrated Fable 5 or Mythos 5 into workflows will need to switch to alternative Anthropic models while access remains suspended. Anthropic has pledged to work to restore access and says the suspension reflects a misunderstanding of the issue.

Source: Anthropic — Statement on US government directive, 12 June 2026 →

Published 9 June 2026 — Microsoft Newsroom, news.microsoft.com

Microsoft and Australian Government Sign First-of-Its-Kind MOU to Strengthen Digital Resilience and Secure AI Adoption

Microsoft and the Australian Government have formalised a Memorandum of Understanding (MOU) to strengthen Australia’s digital resilience in the AI era. The agreement was signed in Canberra by Minister for Home Affairs and Cyber Security Tony Burke and Microsoft’s President of Global Affairs Lisa Monaco. Focus areas include reinforcing the resilience of critical connectivity networks and data centre infrastructure, enhancing cybersecurity threat information-sharing, and supporting the secure adoption of AI. The MOU takes effect immediately and establishes an operating framework for ongoing strategic dialogue to monitor the evolving threat landscape.

Jane Livesey, President of Microsoft Australia and New Zealand, stated that “resilience is a growth strategy” and that the partnership aims to provide the trusted, sovereign digital foundations Australia needs to thrive. For Australian businesses — particularly those operating in critical infrastructure sectors or relying on Microsoft cloud services — the agreement signals a formal commitment to localised infrastructure investment, government-aligned AI governance, and improved cybersecurity co-operation between the public and private sectors.

Source: Microsoft Newsroom — Microsoft and Australian Government deepen collaboration on digital and economic resilience, 9 June 2026 →

Published 9 June 2026 — Microsoft Newsroom, news.microsoft.com

BHP Uses Microsoft Discovery AI to Screen 500,000 Molecules for Copper Recovery

Australian mining company BHP has partnered with Microsoft and computational chemistry firm Prescience Insilico to deploy Microsoft Discovery — Microsoft’s Azure-based agentic AI research platform — to accelerate the search for new copper leaching reagents. The project screened over 500,000 chemical reagents and ran tens of thousands of quantum chemistry simulations to narrow candidates to the most promising molecules for physical laboratory testing in Australia. Microsoft Discovery deploys specialised AI agents across the R&D pipeline, handling scientific literature reviews, hypothesis generation, molecular simulations, and iterative candidate refinement, compressing a research process that traditionally spans decades.

BHP Vice President Innovation Jessica Farrell said the project is “about giving our scientists excellent tools to focus on the most promising copper leaching solutions, sooner.” For Australian businesses in mining, resources, and manufacturing, this case demonstrates how Azure AI platforms can be applied to complex R&D challenges — reducing time to discovery while freeing domain experts for higher-value work. As demand for copper rises due to the energy transition and lower-grade deposits become more prevalent, AI-driven innovation in extraction processes is a direct competitive advantage for Australia’s resources sector.

Source: Microsoft Newsroom, 9 June 2026 →

Published 9 June 2026 — Anthropic, anthropic.com

Anthropic Releases Claude Fable 5: State-of-the-Art Model Now Generally Available

Anthropic released Claude Fable 5 on 9 June 2026, describing it as “a Mythos-class model made safe for general use” that achieves state-of-the-art performance on nearly all tested benchmarks. Stripe reported Fable 5 “compressed months of engineering into days,” completing a full Ruby codebase migration in one day that would have taken two months manually. The model scores highest on Cognition’s FrontierCode evaluation among frontier models at medium effort, highest on Hebbia’s Finance Benchmark for senior-level reasoning, and in internal drug design testing accelerated aspects of the process approximately ten times — with nine of 14 protein targets yielding strong drug candidates. Fable 5 is priced at USD 10 per million input tokens and USD 50 per million output tokens, less than half the previous Mythos Preview price, and is available from today on the Claude API, Amazon Bedrock, Vertex AI, and Microsoft Foundry. Pro, Max, Team, and seat-based Enterprise subscribers can access Fable 5 at no extra cost through 22 June 2026.

For Australian businesses, the release extends Anthropic’s existing Australian footprint. Xero and Canva — which signed multi-year Anthropic partnerships at the Sydney office opening in April 2026 — will integrate Fable 5 capabilities into their platforms, meaning Australian businesses already using Xero or Canva will benefit without a separate subscription. Major consulting firms active in Australia, including KPMG, Deloitte, PwC, and Accenture, are deploying Claude at scale in their advisory and delivery work; Australian clients of these firms will increasingly encounter Fable 5-powered tools in transformation, compliance, and technology engagements. Businesses with direct API access can migrate today using the model identifier claude-fable-5. External red-team testing found no universal jailbreaks in over 1,000 hours of testing, and Anthropic’s classifiers block offensive cybersecurity tasks entirely.

Source: Anthropic, 9 June 2026 →

Published 4 June 2026 — IAPP, iapp.org

2026 Privacy Survey: 93% of Australians Oppose AI Training on Their Data as Trust in AI Companies Falls to 4%

The OAIC’s 2026 Australian Community Attitudes to Privacy Survey (ACAPS) reveals an acute trust crisis. Only 4% of Australians trust AI companies to handle personal data responsibly, and 93% oppose their personal data being used to train AI models without permission. Privacy complaints to regulators have climbed 73% year-to-date, while 65% report that consent to data use feels mandatory rather than genuinely voluntary. Overall, 87% of Australians say their privacy concerns have grown over the past five years, and only 10% believe organisations currently handle their data fairly.

Analysis by John Pane, Chair of Electronic Frontiers Australia, writing for IAPP, argues the findings expose structural risk for Australian organisations deploying AI. Businesses relying on broad consent notices to justify AI training or secondary data uses are increasingly exposed to enforcement action as community expectations sharpen and regulator caseloads grow. Pane calls for Privacy Act reforms mandating lifecycle data controls and prohibiting opaque profiling — changes that, if enacted, would require Australian businesses to audit current AI data practices and implement privacy-by-design approaches.

Source: IAPP, 4 June 2026 →

Published 4 June 2026 — Microsoft Newsroom, news.microsoft.com

Microsoft ANZ: Australia’s Six-Decade Productivity Stall Makes Field AI the Next Major Opportunity

An article by Steven Miller, Area Vice President of Microsoft Australia and New Zealand, published on Microsoft Newsroom on 4 June 2026, identifies field workers as an underserved frontier for AI productivity. Australia’s labour productivity growth is at its lowest in six decades — significantly below the OECD average — and construction industry productivity grew just 17% over the last 30 years compared with 64% across the broader market sector average. With approximately two million skilled tradespeople in field-based roles across Australia, Miller argues that few Australian companies have a concrete plan to deploy AI in this segment of the workforce, despite its scale.

Three strategic priorities are identified for Australian businesses: developing practical, industry-specific AI training for field workers rather than generic programs; involving workers in AI rollout decisions to drive genuine adoption and trust; and sharing success stories to build confidence among tradespeople. The article notes that AI deployed in field settings should reduce administrative burdens — enabling supervisors to spend less time on paperwork and more on skilled work — but warns that implementations that override existing trade expertise, rather than augmenting it, are unlikely to succeed.

Source: Microsoft Newsroom, 4 June 2026 →

Published 3 June 2026 — Anthropic, anthropic.com

Anthropic Launches $100 Million Partner Program as 40,000 Firms Deploy Claude in Enterprise

Anthropic has launched the Services Track and Partner Hub of the Claude Partner Network, backed by a USD 100 million investment in partner training, technical support, and co-marketing. More than 40,000 firms have applied to join the network and over 10,000 consultants have earned Claude certification. Major professional services firms are now deploying Claude at significant scale: Deloitte (470,000 people), Cognizant (approximately 350,000 associates), KPMG (276,000-person workforce), PwC (hundreds of thousands globally), and Accenture (30,000 professionals in training). The program introduces three partner tiers — Select, Preferred, and Global Premier — each with defined certification and customer deployment thresholds.

For Australian businesses, the rollout is directly relevant because the major consulting and professional services firms serving Australian organisations — including Deloitte Australia, KPMG Australia, PwC Australia, and Accenture — are embedding Claude into their advisory and delivery work at scale. Australian companies engaging these firms for technology strategy, transformation programmes, or compliance work are increasingly likely to encounter Claude-powered tools and workflows.

Source: Anthropic, 3 June 2026 →

Published 3 June 2026 — Anthropic, anthropic.com

Anthropic Research: AI Used by 67% of Tracked Cyber Attackers — and Sophistication Gaps Are Closing

Anthropic’s Frontier Red Team published findings from analysing 832 banned accounts involved in malicious cyber activity between March 2025 and March 2026, mapped against the MITRE ATT&CK framework. Key findings: 67.3% of threat actors used AI to write malware; 6.5% deployed it for lateral movement inside networks; and the share of medium-to-high risk actors grew from 33% to 56% over the year — a 1.7× increase. Critically, attacker skill level no longer predicts attack complexity: across all skill levels, threat actors deployed an average of 16–20 techniques, meaning less-skilled attackers can now execute sophisticated post-compromise operations with AI assistance.

For Australian businesses, the research has direct implications for how cyber risk is assessed. Traditional threat models that equate simpler attackers with lower risk are becoming unreliable. Attackers are progressively moving AI assistance deeper into the attack lifecycle — away from initial phishing (which decreased 8.6%) and toward account discovery and lateral movement (which increased 8.9%) — with autonomous, multi-stage attack chains requiring less human involvement. Anthropic has flagged the need to update the MITRE ATT&CK framework to include categories for AI-driven autonomous orchestration, and is engaging MITRE directly on new standards.

Source: Anthropic — AI-Enabled Cyber Threats Research, 3 June 2026 →

Published 1 June 2026 — Anthropic, anthropic.com

Anthropic Files Confidential Draft S-1 with SEC, Opening Path to IPO

Anthropic announced on 1 June 2026 that it has confidentially submitted a draft Form S-1 registration statement to the US Securities and Exchange Commission, taking a formal step toward a potential initial public offering. The filing gives the company the option to go public following SEC review, with share quantities and pricing not yet determined. The announcement was made under Securities Act Rule 135.

For Australian businesses evaluating AI vendor relationships, the development is relevant context. Anthropic opened its Sydney office in April 2026 with Xero, Canva, and Commonwealth Bank as named enterprise partners, and signed an AI safety MOU with the Australian Government in March 2026. A public listing would subject Anthropic to ongoing financial disclosure obligations, giving organisations with multi-year commitments to Anthropic-powered tools — through Xero, Canva, or direct API access — greater visibility into the company’s long-term financial health and stability.

Source: Anthropic — Confidential Draft S-1 Filing Announcement, 1 June 2026 →

Published 28 May 2026 — Anthropic, anthropic.com

Anthropic Raises USD 65 Billion in Series H, Continuing Expansion into Australia

Anthropic raised USD 65 billion in its Series H funding round on 28 May 2026, bringing its post-money valuation to USD 965 billion. The round was led by Altimeter Capital, Dragoneer, Greenoaks, and Sequoia Capital, and includes USD 15 billion in previously committed investments from infrastructure partners including Micron, Samsung, and SK hynix. Funds will be used to advance safety and interpretability research, expand computing infrastructure, and scale products and partnerships globally.

For Australian businesses, the context matters: Anthropic opened its Sydney office in April 2026 with Xero, Canva, and Commonwealth Bank as named commercial partners. This scale of investment signals continued long-term development of the AI capacity underpinning those Australian products. Businesses already accessing Claude through Xero or Canva can expect those integrations to continue to grow.

Source: Anthropic — Series H Funding Announcement, 28 May 2026 →

Published 28 May 2026 — Anthropic, anthropic.com

Anthropic Releases Claude Opus 4.8 with Stronger Reasoning and More Reliable Code Review

Anthropic released Claude Opus 4.8 on 28 May 2026, priced at USD 5 per million input tokens and USD 25 per million output tokens — unchanged from its predecessor — with fast mode now three times cheaper than before (USD 10 and USD 50 respectively). The model shows significant improvements in agentic reasoning, tool-calling accuracy, and browser-based task execution, achieving 84% on the Online-Mind2Web benchmark. A headline improvement: Opus 4.8 is around four times less likely than its predecessor to allow code flaws to pass without flagging them — directly relevant for businesses using Claude for software development or code review.

For Australian businesses using Claude through Xero, Canva, or the API directly, Opus 4.8 is now the model powering those products. Practical improvements include stronger citation precision for financial documents, enhanced analysis of diagrams and PDFs, and the ability to orchestrate hundreds of parallel subagents for large-scale tasks in Claude Code.

Source: Anthropic — Claude Opus 4.8, 28 May 2026 →

Published 19 May 2026 — Anthropic, anthropic.com

KPMG Deploys Claude Across Its 276,000-Person Global Workforce

KPMG announced a global strategic alliance with Anthropic on 19 May 2026, giving all 276,000 employees across 138 countries access to Claude. The deployment is built into KPMG’s Digital Gateway platform — the firm’s primary client-facing tool — with Claude embedded in tax and legal advisory services, private equity portfolio work, internal cybersecurity operations, and enterprise function transformation. According to KPMG’s Vice Chair of Tax, building an AI agent to help clients adjust to changing tax regulations previously took weeks; it now takes minutes.

KPMG has a major presence in Australia as one of the Big 4 professional services firms. Australian businesses that engage KPMG for audit, tax, advisory, or technology services can expect Claude to be embedded in the tools and workflows their KPMG teams use. For organisations in private equity, financial services, or those undergoing technology transformation, this partnership signals the direction large professional services firms are moving with AI-assisted client work.

Source: Anthropic — KPMG Strategic Alliance, 19 May 2026 →

Published 23 April 2026 — Microsoft Newsroom, news.microsoft.com

Microsoft Pledges AI Skills for Three Million Australians by 2028

Microsoft has committed to delivering AI skills to three million Australians by end of 2028 — triple its previous pledge of one million, which was achieved ahead of schedule. The commitment sits within the company’s broader A$25 billion investment in Australian digital infrastructure by 2029. LinkedIn data cited in the announcement shows AI has created 1.3 million new roles in Australia since 2023, with 37% of working-age Australians already using generative AI tools.

Named employer partners include Telstra, Wesfarmers, and Westpac — with 150,000 workers trained through these partnerships in the past year. New initiatives include Microsoft Elevate for Educators, targeting Australian schools. Eight in ten C-suite leaders surveyed now say they prioritise AI confidence over experience when hiring, a direct signal for Australian businesses assessing workforce development priorities.

Source: Microsoft Newsroom, 23 April 2026 →

Published 27 April 2026 — Anthropic, anthropic.com

Anthropic Opens Sydney Office with Xero and Canva as Launch Partners

Anthropic has officially opened its Sydney office and appointed Theo Hourmouzis — previously Senior Vice President at Snowflake, with more than 20 years of Asia-Pacific technology experience — as inaugural General Manager for Australia and New Zealand. The opening follows the Australian Government MOU signed in March 2026 and makes Sydney Anthropic’s fourth Asia-Pacific location after Tokyo, Bengaluru, and Seoul.

Two major Australian software platforms confirmed commercial partnerships at launch: Xero has signed a multi-year deal embedding Claude across its platform, and Canva is integrating Claude via Anthropic Labs. Enterprise customers Commonwealth Bank and Quantium are also named partners. For Australian businesses already using Xero or Canva, Claude-powered features are expected to roll out through these existing products without requiring a separate AI subscription.

Source: Anthropic, 27 April 2026 →

Published 31 March 2026 — Anthropic, anthropic.com

Australian Government Signs AI Safety Agreement with Anthropic

The Australian Government and Anthropic — maker of the Claude AI — signed a Memorandum of Understanding on 31 March 2026, with CEO Dario Amodei meeting Prime Minister Anthony Albanese in Canberra to formalise the agreement. The MOU commits Anthropic to collaborating with Australia’s AI Safety Institute, sharing findings on model capabilities and risks, and supporting joint safety evaluations of advanced AI systems.

As part of the agreement, Anthropic will distribute AUD $3 million in Claude API credits to four Australian research institutions — the Australian National University, Murdoch Children’s Research Institute, the Garvan Institute of Medical Research, and Curtin University — supporting work in genomics, precision medicine, and computing education. A further USD $50,000 in credits will be available to eligible Australian deep tech startups working on drug discovery, climate modelling, and medical diagnostics. Anthropic also confirmed plans to open a Sydney office as part of its Asia-Pacific expansion.

Source: Anthropic — Australian Government MOU Announcement, 31 March 2026 →

Published 22–23 April 2026 — Microsoft Newsroom, news.microsoft.com

Microsoft Commits A$25 Billion to Australian AI Infrastructure

Microsoft Chairman and CEO Satya Nadella visited Australia on 22–23 April 2026, announcing the company’s largest-ever investment in Australia: A$25 billion by 2029 to expand AI and cloud infrastructure, strengthen cybersecurity, and deliver AI skills at national scale. The announcement was made alongside Prime Minister Anthony Albanese.

The investment includes a planned 140% expansion of Microsoft’s existing Australian data centre footprint — bringing capacity across Azure’s three Australian regions significantly higher by the end of 2029. Microsoft also committed to skilling three million Australians by 2028, building on a prior A$5 billion investment from October 2023 that established 29 data centre sites and the Microsoft-ASD Cyber-Shield.

For Australian businesses, the practical implication is substantially more local AI computing capacity — important for organisations with data sovereignty requirements or compliance obligations that restrict data leaving Australian borders.

Source: Microsoft Newsroom — Investing in Australia’s AI Future, April 2026 →

Announced 17 December 2025, launched early 2026 — Gadens / IAPP

Australia Launches AI Safety Institute and Releases National AI Plan

The Australian Government announced its National AI Plan on 2 December 2025, followed by the launch of the Australian AI Safety Institute (AISI) in early 2026. The AISI received AUD 29.9 million in funding and serves as a whole-of-government hub to monitor emerging AI technologies, assess risks, and provide guidance for businesses on responsible AI adoption.

Notably, the government chose not to introduce mandatory AI-specific legislation — instead relying on existing legal frameworks while the AISI develops guidance and testing capability. The National AI Plan’s three core goals are to capture opportunity, spread benefits to all Australians including regional areas, and keep Australians safe. Over $460 million has been committed across research grants, graduate programs, and SME adoption support.

The plan also noted a significant AI adoption gap between regional and metropolitan Australia — 29% versus 40% respectively — which the government has flagged as a priority to address.

Source: Gadens — Australia Launches AI Safety Institute and National AI Plan →

Effective 15 June 2026 — Digital Transformation Agency, dta.gov.au

Australian Government Mandates AI Governance Rules for All Agencies

The first mandatory requirements for government AI use came into effect on 15 June 2026, with all remaining requirements following in December 2026. Under the new framework, every Australian government agency is required to appoint a Chief AI Officer, and a new legal framework governs AI-assisted decisions that require human involvement.

While these requirements currently apply to government agencies rather than private businesses, they signal the direction of travel for AI governance in Australia. Organisations that operate alongside government — as vendors, suppliers, or service providers — should expect their government clients to be asking more questions about how AI is used in their operations from mid-2026 onwards.

Source: Digital Transformation Agency — AI Policy Update, dta.gov.au →