Sponsored by
Welcome back to The AI Field.
Anthropic just launched Claude Science and said it will start developing its own drugs. Alibaba is banning Claude Code after a Reddit post exposed a hidden tracking experiment inside it. Sam Altman wants Washington inside the AI industry, with a reported 5% stake in OpenAI on the table. And the WSJ says SpaceX showed investors a secret xAI phone that Musk swears doesn't exist.
In today's AI Brief:
🧪 Anthropic launches Claude Science and gets into drugs
🚨 Alibaba bans Claude Code over hidden tracking
🏛 Altman offers Washington rules and a 5% stake
📱 The xAI phone Musk says doesn't exist
Read time: 5 minutes.
The best voice models, now across all channels
Most CX platforms do not own the voice. They orchestrate a workflow, then call a third party for speech and transcription. Every hop adds latency, cost, and another vendor to manage.
ElevenAgents is the opposite. They make the voice models the market builds on, and ElevenAgents puts full orchestration on top. Voice, transcription, text-based chat, and reasoning run in one vertically integrated pipeline, so responses come back in <400 milliseconds and sound human, not synthetic.
Plus, you keep full control. Plug in any LLM, integrate tools, webhooks, and MCP servers, and ground responses in your knowledge base. Get an agent live in minutes, then A/B test with Experiments, enforce Guardrails, and version every change.
The payoff: more human conversations, lower latency, and far less time stitching infrastructure together. You build on the models you already trust. Pricing is transparent and flat at $0.08 per minute.
The AI Field: Anthropic released Claude Science, an AI workbench that does for researchers what Claude Code does for developers. The bigger news buried in the launch: Anthropic is entering drug discovery itself, starting with diseases the pharma industry ignores.
Key details:
Claude Science is a customizable app that plugs into the databases, file formats, and analysis tools researchers already use, runs autonomously on high-level instructions, and produces auditable artifacts.
Anthropic will run its own drug discovery programs, initially targeting "neglected" diseases, per life sciences head Eric Kauderer-Abrams.
As a public benefit company, Anthropic says it "can choose programs on patient benefit, including work the commercial market overlooks."
Why This Matters: An AI lab moving from selling tools to owning the science is a different company. If Claude can run the discovery loop itself, pharma stops being a customer and becomes a competitor. Watch which lab follows first.
The AI Field: Alibaba is banning employees from using Claude Code starting July 10, labeling it high-risk software. The trigger: a viral Reddit post showing Anthropic quietly shipped code that could identify Chinese users, which Anthropic now admits was a real experiment. Last week's distillation fight just became a full divorce.
Key details:
The ban starts July 10, and staff are being moved to Alibaba's own coding tool, Qoder.
A Reddit post revealed Claude Code could secretly flag Chinese users. Anthropic's Thariq Shihipar confirmed it was "an experiment we launched in March" against resellers and distillation.
Shihipar says stronger mitigations have landed since, and the team had "been meaning to take this down for a while."
Why This Matters: Anthropic accused Alibaba of distilling Claude just days ago, and now we know it was also watching who used its coding tool. Every company running AI coding agents just got a reminder that the tool can see you back. Expect more bans, in both directions.
The AI Field: After a June of crackdowns, Sam Altman used an FT op-ed to call for a US-led forum with real power to set AI safety standards and decide who gets the most advanced models. A separate FT report says OpenAI floated handing the US government a 5% stake.
Key details:
Altman pitched an IAEA-style referee for AI, pointing to atomic energy, aviation, and banking as proof it can work.
OpenAI reportedly discussed a 5% government stake, plus a dividend fund other US labs would pay into.
The op-ed came out of June's G7 summit in France, where AI executives sat down with heads of government.
Why This Matters: The labs have realized they can't outrun regulation, so Altman is trying to design it himself. A government that owns 5% of OpenAI is a government with a reason to protect OpenAI. That's not safety policy, that's a moat.
The AI Field: The Wall Street Journal reports SpaceX demoed a prototype xAI-powered phone to IPO investors. Musk's response on X: "utterly false." Someone is wrong, and both options are a story.
Key details:
The WSJ says the prototype was shown to investors as part of SpaceX's post-IPO pitch.
Musk publicly denied the report on X, calling the claim "utterly false."
SpaceX went public in June at a $1.77 trillion valuation with xAI inside, and bought Cursor for $60B days later.
Why This Matters: An xAI phone with Starlink connectivity would be the first real hardware threat to the iPhone in a decade, and Musk denials have a track record of aging badly. Watch what SpaceX files, not what Musk tweets.
🔥 THE AI FIELD'S TAKE
The trust wars are loud this week. Alibaba doesn't trust Anthropic's code. Washington doesn't trust the labs. Investors aren't sure they can trust Musk's denials. Altman's fix is to hand the referee a jersey: a forum designed by OpenAI, and a government that owns a slice of the winner. Clever. Just don't confuse it with neutrality.
And while everyone fights over who to trust, Anthropic pointed Claude at something nobody argues with: curing the diseases the market forgot. It's the best PR move in AI right now, precisely because it's more than a PR move.
In 2026, the scarcest resource in AI isn't compute. It's trust. Anthropic just found a new way to earn it.
Nvidia's next-gen AI rack is reportedly delayed a year. The Kyber architecture built for Rubin Ultra chips slips to 2028, and NVDA swung premarket on the report.
Zuckerberg told Meta staff AI agents are behind schedule. When the man spending the most on agents says they haven't progressed as fast as he hoped, believe him.
Microsoft launched Frontier Company. A $2.5B plan to embed 6,000 in-house engineers at client sites to build and run AI systems.
A tiny custom model just beat every frontier model. Thinking Machines and Bridgewater trained an open model on the fund's own judgment: 84.7% on its tasks, at 13.8x less cost.
Anthropic reportedly approached Samsung about a custom AI chip. It comes right after poaching Clive Chan from OpenAI's Jalapeño team.
China's Kling AI raised $2B. The video AI is going global after OpenAI shut down Sora.
👀 WHAT TO WATCH
Alibaba's Claude Code ban: takes effect July 10. Watch whether other Chinese tech giants follow within the week.
Fable 5 usage limits: the temporary 50% cap eases after July 7, then moves to usage credits.
GPT-5.6 general availability: still limited to ~20 orgs after the White House slow-roll. OpenAI says broad access is "coming weeks."
Be the one who already read it. TLDR is the 5-minute tech brief that 7M+ subscribers start their morning with, curated by ex-Google and Anthropic engineers.
WRAP-UP
That's all for today's AI Brief.
🧪 Anthropic launches Claude Science and starts developing its own drugs.
🚨 Alibaba bans Claude Code after a hidden tracking experiment comes to light.
🏛 Altman pitches a US-led AI regulator, with a 5% OpenAI stake reportedly on the table.
📱 The WSJ says SpaceX showed investors an xAI phone. Musk says it never happened.
What did you think of today’s edition?
Until next time!
Olle | Founder of The AI Field