Anthropic calls for global AI development freeze amid safety concerns
Consensus Summary
Anthropic, the world’s most valuable AI company, called for a global freeze on advanced AI development on June 5, 2026, warning that humanity risks losing control of the technology as it approaches 'recursive self-improvement,' where AI could autonomously design its successors. The company cited internal data showing over 80% of its code is now written by its own AI, Claude, and highlighted risks like cyberattacks, with its most powerful model, Mythos, withheld from public release due to safety concerns. The Australian government, led by Assistant Minister Andrew Charlton, welcomed the call for a more cautious approach, though it contrasts with its December 2025 decision to abandon mandatory guardrails for high-risk AI in favor of a lighter-touch regulatory model. Polling shows 68% of Australians fear losing control to AI, while groups like PauseAI Australia argue that any freeze should be governed by international treaties rather than industry alone. The ABC notes that Anthropic’s call comes amid a resurgent AI stock market bubble and its own IPO plans, while the Guardian highlights tensions between Anthropic’s safety rhetoric and its reported collaboration with the NSA on offensive cyber operations.
✓ Verified by 2+ sources
Key details reported by multiple sources:
- Anthropic, the company behind Claude chatbot, called for a global freeze on the development of the most powerful AI systems on June 5, 2026.
- Anthropic's CEO Dario Amodei, who left OpenAI over safety concerns, stated there is a 25% chance that 'things go really, really badly' with AI.
- Anthropic's Mythos AI model was withheld from public release due to fears it could be used for devastating cyberattacks, and was released in a staggered rollout to fix security flaws.
- Anthropic's internal data shows that as of May 2026, more than 80% of the code merged into its codebase was authored by Claude.
- The Albanese government in Australia welcomed Anthropic’s call for a more considered approach to frontier AI development, with Assistant Minister for Science and Technology Andrew Charlton stating the pace of progress 'must not outstrip understanding or undermine safety across society'.
- Anthropic’s call for restraint contrasts with Australia’s federal government’s decision in December 2025 to shelve plans for mandatory guardrails on high-risk AI, opting for a lighter-touch model.
- Australia’s AI Safety Institute is expected to open within months, and it signed a memorandum of understanding with Britain’s AI Security Institute last week.
- Polling from EY found that 68% of Australians are worried about losing control over decisions made by AI.
- Anthropic’s blog post titled 'When AI builds itself' discusses the concept of 'recursive self-improvement', where AI systems could design and build their own successors with little human input.
Points of Difference
Details reported by only one source:
- PauseAI Australia, part of an international movement, welcomed Anthropic’s intervention but said the terms of any freeze should not be set by industry, advocating for a treaty negotiated by governments.
- Former industry minister Ed Husic, who was dropped from the ministry last year, had said a piecemeal approach to AI regulation would leave dangerous gaps.
- Anthropic’s Mythos model was evaluated by Britain’s AI Security Institute, which was among the first to publicly assess an earlier version of the model.
- No additional unique details beyond theage and guardian.
- Anthropic’s Mythos model was confined to a select group of insiders under Project Glasswing, including a small number of Australian firms, to prepare defenses against potential misuse by non-democratic actors.
- The stock market AI bubble surged again on April 7, 2026, eight days after Anthropic announced Mythos, with global share prices up 20% and AI stocks up 32% since March 30.
- China became the first country to regulate emotional attachment to humanoid AI robots through 'Interim Measures for the Administration of Anthropomorphic AI Interaction Services', prohibiting excessive emotional dependency.
- Anthropic’s IPO filing could value the company at $1 trillion, and the company is part of a new wave of trillion-dollar AI businesses alongside OpenAI and SpaceX.
- The AI industry is expected to develop into three layers: GPU vendors, cloud providers, and LLM APIs, with higher margins in vertical products embedding LLMs.
- Anthropic has embedded engineers inside the National Security Agency (NSA) to help use its Mythos model for offensive cybersecurity operations, despite a legal battle with the Pentagon.
- Steven Murdoch, a professor at University College London, called Anthropic’s Mythos announcement a 'marketing post' and said the company’s definition of AI safety is narrow, supporting US offensive capabilities.
- Heidy Khlaaf, chief AI scientist at the AI Now Institute, also suggested there could be more hype than substance behind Mythos’s capabilities.
Contradictions
Conflicting information between sources:
- The Guardian suggests Anthropic’s call for a pause is not new and echoes past proposals, while theage and smh frame it as a significant new intervention.
- The ABC describes Mythos as 'the most dangerous AI model yet,' while the Guardian suggests its capabilities may be overstated and called it a 'marketing post'.
- Theage and smh state that the Albanese government welcomed Anthropic’s call, but the ABC does not explicitly mention government response beyond the initial statement.
- The Guardian implies that Anthropic’s support for NSA cyber operations contradicts its call for a pause, while other sources do not mention this contradiction.
Source Articles
Anthropic urges global freeze on AI as it warns of losing control
The tech giant has offered to halt work on more powerful systems on one condition: that its rivals in the US and China agree to stop at the same time.
Anthropic urges global freeze on AI as it warns of losing control
The tech giant has offered to halt work on more powerful systems on one condition: that its rivals in the US and China agree to stop at the same time.
Anthropic wants caution but the market wants more AI
Anthropic wants caution as AI intelligence surges but the stock market can't get enough.
Anthropic urges AI development ‘pause’ and conversation about risks
US firm says it will convene policymakers, in post detailing progress of its Claude model towards ‘recursive self-improvement’ Anthropic has floated the idea of a worldwide “temporary pause” on AI development – and said it was going to convene “policymakers” to discuss the dangers of advanced AI – in its latest release touting the capabilities of its products. In a long post on Thursday, Anthropic detailed the progress of its AI model, Claude, towards “recursive self-improvement” – that is, bein