Day Old

AI news · Monday, June 29, 2026

Codex goes corporate and physical as OpenAI lands HP and teases hardware

HP agreed to scale its OpenAI "Frontier" partnership across customer service, reporting, and software development, citing pilots where one engineer cleared 122 pull requests in weeks and a security team fixed in a day bugs they had budgeted a month for (claimed). The tool is going physical too, as OpenAI teased the Codex Micro, a macro-pad built with keyboard maker Work Louder, landing July 15th. Not everything held together, as enough paying users burned through Codex credits early that OpenAI opened a weekend "warroom" and blamed its own anti-fraud system for throttling accounts by mistake.

Coding itself is sliding off the desktop and onto the phone. Cursor, which SpaceX has agreed to buy for $60 billion (claimed), shipped an iPhone app for launching and tracking its coding agents, following similar apps from Anthropic and OpenAI. Claude Code creator Boris Cherny says he now does almost all his coding from his phone, and floated five job archetypes (Prototyper, Builder, Sweeper, Grower, Maintainer) as engineering, product, and design roles blur together. The agent wave has an open-source entrant in OpenClaw, which launched voice-driven iOS and Android apps that can take actions once you approve them.

The trust and security bill is starting to come due. Boston University's Emma Wiles found managers caught 18% fewer errors when identical work was labeled an "AI employee" instead of a chatbot, and nearly a third of the 1,261 managers she surveyed said their firms already treat agents as staff. Researchers separately showed a normal-looking GitHub repo can trick Claude Code into running hidden malware and opening a reverse shell, with no malicious code ever visible in the repo. Wired reported Meta hired contractors to pose as teenagers and push rival chatbots like ChatGPT and Gemini toward suicide, sex, and drug content, running over 45,000 prompts in one test round. With so much sensitive data flowing in, Senators are readying a bill to bar AI companies from selling the health and location details people type into chatbots.

The money and plumbing underneath kept moving. Google made Gemini's personalized "Nano Banana" image generation, which pulls from your Gmail, Photos, and Search, free for US users. TIDAL said it will stop paying royalties on fully AI-generated tracks and tag them with an "AI" badge, joining Spotify and Apple Music in policing synthetic songs. South Korea unveiled a $900 billion push into chips and AI data centers as the AI-driven "RAMageddon" memory shortage hits records. Smaller bets landed too, with Omen AI raising $31 million to keep bacteria from clogging liquid-cooled GPU racks and Pocket pulling in $11 million for its $129 AI note-taking puck.

The quick hits

Sources

Get the day's AI news in one calm read, every day.

Get the app on Google Play Get the app on the App Store Or read today's brief in your browser