1. The week two AI coding vendors moved the paywall on solo developers Within seven days, GitHub and Anthropic redrew the boundaries of what individual developers get for a flat monthly fee. GitHub published changes to its Copilot individual plans.

2. The pitch-deck line that used to be embarrassing: we spend more on AI than on people According to 404 Media, a new class of startup founders is turning a once-awkward admission into a public selling point.

3. MIT Put Weaponized Deepfakes on Its 2026 Watchlist. OpenAI Spent the Same Week Shipping Both Sides. MIT Technology Review named weaponized deepfakes one of its ten breakthrough threats for 2026 on April 21.

In Brief

• Google launched TPU 8t and 8i at Cloud Next Google unveiled two eighth-generation TPUs pitched as faster and cheaper than the prior line, aimed at agentic workloads. Nvidia GB300 chips still run inside Google Cloud alongside them, so customers pick either stack.

• Thinking Machines Lab signed a multibillion-dollar deal with Google Cloud Mira Murati's lab will train on Google Cloud infrastructure powered by Nvidia GB300 chips. The contract deepens an existing relationship and gives Google a second flagship AI tenant after Anthropic.

• Unauthorized users accessed Anthropic's Mythos model Bloomberg reports a small group obtained access to Mythos, the cybersecurity model Anthropic previously restricted over dual-use risk. One member is identified as a third-party contractor; the group coordinates through a private online forum.

• OpenAI opened workspace agents to Business, Enterprise, Edu, and Teachers plans Teams can build cloud agents that run tasks on their own, including a product-feedback scraper that posts reports to Slack and a sales agent that works pipeline data. Availability is limited to paying team tiers.

• Elizabeth Warren called the AI economy a bubble The senator told a Vanderbilt Policy Accelerator audience that AI investment patterns show parallels to the run-up to 2008 and an AI failure could trigger the next financial crisis. Warren led creation of the CFPB after the last one.

• OpenAI signed Infosys to resell its tools into enterprise accounts The integration targets software engineering, legacy modernization, and DevOps at Infosys clients. OpenAI now reaches enterprises through a systems integrator rather than direct sales only.

• Google put Gemini "auto browse" into Chrome Enterprise Workers on managed Chrome deployments can hand research, data entry, and similar browser tasks to Gemini and let the model complete them. The feature ships as part of enterprise Chrome rather than a separate install.

• Google aimed Gemini Enterprise Agent Platform at IT users, not business staff The new agent builder assumes a technical audience, a split from Microsoft's Copilot Studio approach that targets line-of-business users. Google is positioning agent rollout as an IT-owned process.

• OpenAI made ChatGPT for Clinicians free to verified US medics Access is open to US physicians, nurse practitioners, and pharmacists who pass credential verification. The product is scoped to clinical care, documentation, and research.

• 10x Science raised $4.8M seed to triage AI-generated drug candidates The startup helps pharma researchers decide which AI-proposed molecules are worth advancing. Generative chemistry pipelines have outpaced validation capacity inside drug companies.

• Sony's Ace robot beat top-ranked ping-pong players Sony AI's Ace combines cameras and learned control to play competitively against elite human competitors, occasionally winning points. Prior robots such as Omron's FORPHEUS only held up against amateurs.

Read the full edition →