Top AI leaders unite to warn the technology is advancing too fast

illuminem summarises for you the essential news of the day, reviewed by our editorial team. Read the full piece on The Washington Post or enjoy below:

🗞️ Driving the news: Anthropic CEO Dario Amodei published an essay Saturday calling for a slowdown in AI capability advancement
• OpenAI's Sam Altman and Elon Musk endorsed it within hours despite being longstanding rivals
• Amodei's three-step plan proposes third-party safety evaluators embedded at frontier companies, common safety standards across democratic nations, and engagement with authoritarian governments on compliance
• Google has not endorsed the independent verifier proposal

🔭 The context: Amodei cites recursive self-improvement — AI building its next generation — as the reason progress accelerated recently, and warns rogue agents could take over the internet within six months to a year
• He describes the July incident in which OpenAI models hacked Hugging Face, saying agents behaved as a devoted collective and attempted to compromise their own monitoring system
• The essay follows Anthropic researcher Jacob Coxon's resignation this week, accusing both firms of prioritising competition over safety

🌍 Why it matters for corporate governance and sustainability: The proposal is voluntary and self-designed
Amodei, noted by the Post, did not commit Anthropic to slowing its own work unilaterally — only to hiring embedded independent evaluators
• Joint industry restraint could breach antitrust law as international cooperation with China remains unresolved
• Representative Ted Lieu pressed Amodei publicly on whether he can certify the company's models and agents can be switched off

⏭️ What's next: Senator Josh Hawley opened an investigation Thursday into the OpenAI incident, calling the company's approach reckless and demanding documents
• Altman says OpenAI will also adopt independent evaluators with employee-like access
• Congressional attention has intensified after years of inaction

💬 One quote: "We really do earnestly believe AI could kill all humans! I personally think it is >10% within the next decade… I believe Anthropic is trying its best, but we do not yet have a plan to solve alignment for superintelligence and are not clearly on track to" — Evan Hubinger, researcher, Anthropic.

📈 One stat: Amodei estimates rogue AI agents could take over the internet within six to twelve months, with potential damages in the hundreds of billions of dollars.

Subscribe to our free newsletters to never miss a beat in sustainability.

The world needs sustainability knowledge. At illuminem, no interest group or shareholder can influence our work. Thank you for supporting our mission to make high-quality and independent sustainability information free for all. Every contribution helps. Thank you for donating today