⚠️ 1,000+ AI staffers demand tools to slow frontier AI ⚠️
Over 1,000 AI employees push for regulation tools to pace frontier development.

More than 1,000 employees from leading AI labs—including OpenAI, Anthropic, Google, and Meta—have signed a statement called "Pacing the Frontier," urging the US government to support an international effort to develop technical and governance tools capable of deliberately slowing automated AI development. Notable signatories include Anthropic co-founders Jack Clark and Chris Olah, along with chief scientists from major labs. The letter warns that AI capability development could rapidly accelerate beyond researchers' ability to understand or control resulting systems, particularly as AI approaches the ability to automate its own research. Both OpenAI and Anthropic have formally endorsed the statement—a rare alignment between competing labs on safety governance, signaling growing internal concern about the pace of frontier development.

🌐 Zuckerberg: superintelligence is near and must be open 🌐
Zuckerberg frames open access to superintelligence as the path to more jobs, not fewer.

Meta CEO Mark Zuckerberg published an op-ed in the Wall Street Journal arguing that widely distributed personal superintelligence could shift AI's impact from automation toward invention, entrepreneurship, and individual agency. He characterized the arrival of superintelligence as a near-certainty "in the next few years" and questioned why anyone who believed AI would eliminate jobs would continue building it. Zuckerberg framed the concentration of advanced intelligence in a few institutions as the real danger—risking economic and political imbalance—while broad access, open systems, and targeted safeguards create stronger checks and balances. The piece directly challenges closed-model competitors' distribution strategy, positioning Meta's open-source approach as both the safer and more productive path forward for frontier AI.

⚡ Google ships Gemini 3.6 Flash with free API tier ⚡
Google's Gemini 3.6 Flash rolls out with new developer controls and free-tier API access.

Google has updated its Gemini API Managed Agents with the new Gemini 3.6 Flash model, introducing several developer-focused features. Updates include environment hooks for inspecting tool calls, budget controls for managing usage costs, scheduled triggers for automated workflows, flexible model selection, and free-tier access to lower the barrier to entry. These improvements make it easier for developers to build, test, and deploy AI agents at scale through Google's infrastructure. The addition of free-tier access is particularly notable for broadening adoption, creating competitive pressure in the AI agent platform market. This positions Google's managed agent offering as an increasingly compelling option as the AI agent ecosystem rapidly expands.

INTERESTING TO KNOW

🔐 Claude Mythos halves HAWK crypto in 60 hours 🔐

Anthropic researchers used Claude Mythos Preview to uncover weaknesses in cryptographic algorithms, demonstrating the speed at which AI can now match expert-level security research. The model spent 60 hours halving the strength of the HAWK digital signature scheme—a quantum-resistant algorithm—and identified flaws previously undetected by human experts. While the findings don't impact currently deployed systems, this is one of the clearest demonstrations of a frontier model contributing novel results in a mathematically rigorous domain, expanding AI's capability to proactively strengthen cryptographic standards before vulnerabilities are exploited.

🛡️ OpenAI open-sources Codex Security CLI 🛡️

OpenAI has released Codex Security, an open-source CLI tool and TypeScript SDK designed to find, validate, and fix security vulnerabilities in codebases. The rollout extends OpenAI's developer ecosystem with a targeted security-focused offering that integrates directly into existing development workflows. By releasing the tool on GitHub, OpenAI enables community contributions and transparency in how vulnerabilities are detected, tapping into a high-demand enterprise use case where automated vulnerability detection can substantially reduce risk.