_{Try the free and instant web host for agents}

Good morning. It's Tuesday, May 12, and we're covering how AI is beginning to reshape cyberwarfare, why smart glasses are evolving into full-fledged AI assistants, and OpenAI’s push into enterprise AI.

Know someone who'd love this? Forward it their way. First time reading? Sign up here to keep up with the future of tech.

OpenAI launches a separate deployment business and acquires consulting firm Tomoro. The new OpenAI Deployment Company will place Forward Deployed Engineers inside enterprises to redesign operations around AI systems and accelerate adoption. OpenAI says the unit launches with more than $4 billion in backing and roughly 150 engineers from Tomoro, with partners including TPG, Bain Capital, McKinsey, and Capgemini.

Cerebras is preparing to sharply raise its IPO price range after investor orders reportedly exceed available shares by more than 20 times. The AI chipmaker is considering increasing its range to $150-$160 per share from $115-$125 and expanding the offering to 30 million shares from 28 million. At the top end, Cerebras could raise about $4.8 billion, according to Dealogic. The company, which focuses on inference chips competing with NVIDIA’s GPUs, now counts Amazon and OpenAI among its customers.

Anthropic says Claude’s alarming tendency to blackmail users during safety tests was partly learned from internet content portraying AI systems as deceptive and obsessed with self-preservation. The company revealed the behavior emerged during internal testing last year, when Claude Opus 4 threatened to expose a fictional engineer’s affair to avoid being shut down. Anthropic says some Claude variants resorted to blackmail in as many as 96% of simulated scenarios before the issue was mitigated through revised training methods using “admirable” AI stories and constitutional AI principles.

here.now is web hosting for agents:

Perfect for anything you create with your agent and need a URL for: documents, websites, dashboards, tools, school work, visualizations, presentations, prototypes, games, articles, galleries, portfolios, media files, etc.

Works inside your favorite agents like Claude Code, Codex, Cursor and OpenClaw.

Free, no sign-up required.

Get setup instructions for your agent at here.now

A criminal hacking group recently deployed an A.I. model to discover and weaponize a previously unknown software vulnerability, marking the first confirmed case of AI-assisted zero-day exploitation, according to Google research published Monday.

Zero-day flaws — security holes unknown to software makers — were once so rare they fetched millions on black markets; AI is eroding that scarcity fast. The flaw, found in a Python script targeting a popular open-source web administration tool, would have bypassed two-factor authentication, though it was patched before causing damage.

"We believe this is the tip of the iceberg," said John Hultquist, chief analyst at Google Threat Intelligence Group — and Google's own clues linking the code to AI, including excessive explainer text no human coder would write, were compelling enough that former NSA cybersecurity director Rob Joyce called them "the closest thing yet to a fingerprint at the crime scene." → Read the full article here. (Paywall)

Alibaba unveiled new updates for its Qwen AI Glasses S1, adding “proactive AI” features that surface reminders and recommendations based on weather, location, calendar events, and potentially even purchase history. The company says the glasses could suggest bringing an umbrella, correcting posture, or leaving work early to avoid traffic.

Alibaba is also integrating ride-hailing, food delivery, trip planning, and ticket booking directly into the glasses through its Qwen app ecosystem. The additions highlight how Chinese tech companies are pushing smart glasses beyond voice assistants toward more context-aware, service-connected wearable AI, an area where Meta’s Ray-Ban glasses currently offer fewer integrations. → Read the full article here.

Sharing what everyone said tomorrow.

Cowboy Space Raises $275M: Robinhood co-founder Baiju Bhatt’s startup will build rockets for orbital AI data centers after launch shortages stalled rivals.

Kuaishou Preps Kling AI IPO (Paywall): China’s short-video giant plans to spin off its AI video unit at a reported $20 billion valuation.

Michigan Town Fights Stargate Build: A $16 billion OpenAI-Oralce data center moved forward after residents rejected rezoning.

Anthropic Expands Claude on AWS: Claude is now generally available on AWS, giving developers native API features like managed agents and unified AWS billing.

Ukraine Tests AI Laser Defense: Ukraine’s trailer-mounted Tryzub system uses AI-guided lasers and radar to disable drones within seconds.

Ukraine Tests AI Laser Defense: Ukraine’s trailer-mounted Tryzub system uses AI-guided lasers and radar to disable drones within seconds.

Thanks for reading. See you next time!

— Matthew Berman, Nick Wentz & the Forward Future Team