A public release two months after a private debut rattled Wall Street, enabled by safeguards that block responses in specific high-risk areas.

A “safe” public model and a partner-only cyber version force buyers and boards to rethink access and risk controls.

A new Claude model is coming out of Anthropic's Mythos cybersecurity hold, with safeguards that limit high-risk outputs.

WhatsApp says it caught NSO spear phishing, including test accounts, in Jordan and Lebanon, after a federal ban.

Apple’s WWDC move removes most manual password change work, but the reliability test is the hard part.

It marks the first reported US sea-rescue using an uncrewed surface drone, with Task Force 59 in the driver’s seat.

Neo4j CEO Emil Eifrem says on-prem, air-gapped deployments are the point, as regulators worry about U.S. access leverage.

The London mayor intervened in late May, halting Palantir's intelligence automation deal and triggering a procurement fight.

CVE-2026-23111 in nf_tables turns a tiny coding error into a use-after-free exploit for unprivileged root access.

Meta says it is widening how partners' data shapes your Reels feed and AI responses, without collecting new data.

Developers can team up on discounted subscription packages, pushing the App Store closer to “streaming-style” bundling.

If Apple uses Siri as an interface layer, executives can bet on distribution and on-device leverage, not model-building.