Enterprise IAM Wasn’t Built for AI Agents — and Agentic Deployments Are Stuck in Pilots Because of It

AI agents running hospital records and factory inspections are generating non-human identities that enterprise security teams cannot inventory, scope, or revoke at machine speed. Cisco President Jeetu Patel identifies identity governance — not model capability or compute — as the structural barrier keeping agentic AI out of production.

Thinking Machines Previews Near-Realtime AI Voice and Video — the End of Turn-Based AI

Thinking Machines has previewed near-realtime AI voice and video conversations using what it calls interaction models — a fundamental departure from the input-then-wait paradigm on which every current AI interface is built.

A Palo Alto Student Was Accused of AI Cheating by a Faulty Detector. Now It's a Civil Rights Lawsuit.

A Palo Alto high school sophomore has filed a civil rights lawsuit against his school after an AI-detection tool incorrectly flagged his work as machine-generated. The case is believed to be among the first in the US to test the legal liability of AI detection systems used in institutional settings.

Google Is Rethinking Laptops Again — the Android-Powered Googlebook Arrives Ahead of I/O

Malware Is Hiding in Google Search Ads — and It Targets People Searching for AI Apps

A malware campaign has been discovered placing trojanised downloads behind Google Search paid ads, intercepting users searching for legitimate AI applications — including Claude's Mac client. The campaign exploits the paid-ad position to redirect victims to credible-looking distribution pages.

Claude Code, Google Antigravity, and Codex Go Head to Head — One Month of Real Deadlines, One Clear Winner

A month-long hands-on comparison of Claude Code, Google Antigravity, and OpenAI's Codex across real development tasks — complex refactoring, sustained agentic workflows, multi-file context — has yielded a clear winner on sustained usefulness under production pressure.

Google Meet's New AI Dubbing Changes Who Can Actually Participate in the Meeting

Google Meet has introduced an AI dubbing feature that translates and dubs meeting participants in near-realtime, allowing speakers of different languages to join and contribute without interpreters or a shared language as a prerequisite.

Texas AG Sues Netflix for Covertly Collecting User Data and Deliberately Designing for Addiction

Texas Attorney General Ken Paxton has sued Netflix, alleging the platform collects personal data without user consent and has deliberately engineered its service to maximise time spent — particularly for children. The suit is brought under Texas data privacy statutes and targets both the data collection and the design patterns.

The Copilot Key Is Now on Enterprise Laptops — Whether IT Ordered It or Not

Microsoft has placed a dedicated Copilot hardware key on new Windows laptops, occupying a position that previously held keyboard shortcuts embedded in employee muscle memory. The key ships in enterprise refresh cycles before most IT governance teams have updated AI access or data-handling policies.

Google Reveals Plans to Rebuild Android Auto With a Cohesive New Design Ahead of I/O

Google has announced plans to overhaul Android Auto with a redesigned, more cohesive user experience — addressing a longstanding design gap versus Apple CarPlay — with the full reveal expected at Google I/O on 19–20 May 2026.