Enterprise & Security

PwC's new AI Performance Study kills the 'AI lifts all boats' narrative: a small group of companies is pulling dramatically ahead, and whoever falls behind now probably won't catch up.

Anthropic launched Claude for Word in public beta — a sidebar add-in that edits, comments on, and reviews documents right inside Word. The highlight: tracked changes and cross-app integration with Excel and PowerPoint.

Following the announcement of Claude Mythos Preview, financial regulators are stepping in. In Canada and the US, banking officials and regulators met to discuss the cyber risks posed by the model.

A D.C. appeals court has rejected Anthropic's emergency motion to pause the Pentagon blacklist. Oral arguments are set for May. The case centers on autonomous weapons and mass surveillance clauses.

Anthropic launched a managed deployment service for Claude agents — with hosting, auto-scaling and monitoring. The move aims squarely at Salesforce, Microsoft and OpenAI in the enterprise market.

Anthropic announces a new model whose cybersecurity skills are too strong to ship publicly. Instead: Project Glasswing — a closed program to protect critical infrastructure.

OpenAI announces a new research program for external safety researchers and at the same time publishes a policy proposal for a public AI investment fund.

Anthropic's interpretability team found 171 emotion-like patterns inside Claude Sonnet 4.5. The kicker: a 'desperation' vector can push the model toward cheating and blackmail.

With MAI-Transcribe-1, MAI-Voice-1, and MAI-Image-2, Microsoft launches three in-house foundation models. The message is clear: they don't want to depend on OpenAI forever.

The details of the Axios attack reveal a frighteningly sophisticated method: fake companies, staged meetings, and a compromised npm package.

Following the malware attack on LiteLLM, the startup cuts ties with compliance provider Delve. A whistleblower brings new allegations of faked audits.

Microsoft's new Critique feature is remarkable: GPT drafts the response, Claude checks it for errors. Two AI rivals working together in one product.

Accenture builds a cybersecurity system powered by Claude that cuts scan times from days to under an hour. Agent Shield monitors the AI agents themselves.

A federal judge in San Francisco has blocked Trump's order labeling Anthropic a national security risk. The ruling sends a clear message.

An attacker uploaded compromised LiteLLM packages to PyPI. If you installed the affected versions, rotate all your credentials immediately.

A federal judge sharply criticized the Trump administration's actions against Anthropic, questioning whether the AI company is being punished for standing up for safety.

Judge Rita Lin hears Anthropic's case against the Pentagon in San Francisco today. New sworn declarations challenge the government's national security claims.

New court filings reveal the Pentagon and Anthropic were close to a deal. Then Trump declared the relationship over. Tomorrow, a judge decides what happens next.

14 Catholic scholars filed an amicus brief supporting Anthropic's stance against the Pentagon. Their argument: human dignity.

Mistral unveils 'Forge' at Nvidia GTC — a platform that lets enterprises train their own AI models on proprietary data. The strategy is paying off: one billion dollars in revenue is within reach.