Enterprise & Security

Anthropic's CEO lays out a comprehensive regulation blueprint: FAA-style testing for AI models, government veto power on dangerous releases, and universal basic income funded by AI taxes.

After massive backlash, Anthropic reverses course: invisible throttling of Claude Fable 5 on AI research queries will be replaced with visible safeguards.

Anthropic signs a multi-year global alliance with IT services giant DXC. Tens of thousands of Claude-certified engineers will bring Claude into the critical systems of banks, airlines, and government agencies.

OpenAI banned a cluster of accounts likely run from China that used ChatGPT for covert influence operations aimed at stoking opposition to AI data centers in the US.

Oracle customers can now use OpenAI's frontier models and Codex through their existing Universal Credits. No new procurement process, no separate contract – a classic distribution move.

A single paragraph buried in Fable 5's 319-page system card is causing a storm: when you ask for help with frontier AI research, the model deliberately weakens its answers — without telling you. Researchers are calling it 'secret sabotage'.

OpenAI is rolling out 'Memory Sources' across all ChatGPT models: for every personalized answer, you can see which memories and past chats fed into it — and correct or delete them. Plus a refresh for GPT-5.5 Instant.

A new security mode against prompt injection: in Lockdown Mode, ChatGPT cuts web access, Agent Mode, Deep Research, and file downloads. An honest move - and an admission at the same time.

Companies are sending simple tasks to cheap models and only the hard stuff to Claude or GPT. It saves millions — but threatens the business models of frontier labs.

832 banned accounts, cleanly mapped onto MITRE ATT&CK: Anthropic's new report shows attackers are pushing AI deeper into the attack lifecycle — and that the established risk signals are going dull.

A 269-page bipartisan bill in Congress would unify AI regulation at the federal level. California's AI bills, Colorado's AI Act — all frozen for three years. What's inside and why it matters.

Anthropic is restructuring its partner program — with three tiers that show how deeply a firm is actually invested in Claude, and a portal that refreshes daily who has delivered what.

Project Glasswing is growing fast: Anthropic's most powerful model now scans critical infrastructure across 15 countries. Over 23,000 vulnerabilities found so far.

Who actually checks whether an AI model is safe? OpenAI puts out a playbook for trustworthy third-party evaluations — and makes its Frontier Governance Framework more transparent.

A company burned through $500 million on Claude in a single month because it never set usage limits. Not an isolated case — and a lesson for anyone rolling out AI at scale.

Anthropic has a model that's too powerful for public release. Mythos remains locked away - for security reasons.

OpenAI shows how its safety practices align with the EU AI Act and California's Transparency Act. A document between compliance and strategy.

After Google's AI overhaul of Search at I/O, users are fleeing to DuckDuckGo. App installs spiked up to 30 percent over six days.

Daniel Stenberg, the maintainer of curl, is sounding the alarm: AI-generated security reports are flooding open-source projects with junk. The bug bounty program was already shut down once because of it.

OpenAI is offering its cybersecurity tools to US election officials and rolling out SynthID watermarks against deepfakes. With the midterms approaching, the stakes are high.