Schwarz Group Funds Europe's Sovereign AI as OpenAI Apologizes and Murati Poaches Meta Engineers

01Lidl's Owner Bankrolled Europe's Sovereign AI Bet the Same Week Washington Weighed China Sanctions

The Trump administration is weighing sanctions on Chinese firms it accuses of "industrial-scale" AI model theft. The penalties could derail an upcoming Trump-Xi summit. Beijing has called the accusation "slander."

The same week, two announcements landed that look less like coincidence than consequence.

In Europe, Canadian startup Cohere is taking over Germany's Aleph Alpha, the country's best-known foundation model lab. The deal is being financed by Schwarz Group, parent of discount retailer Lidl. Both the British and German governments have publicly endorsed the merger. The pitch to European enterprise buyers: a sovereign alternative to American models, backed by a balance sheet that does not depend on US venture capital.

In Tokyo, Anthropic and NEC announced a joint effort to build what they call Japan's largest AI engineering workforce. The arrangement embeds Anthropic's models inside NEC's enterprise channels and stands up local engineers who can deploy Claude under Japanese data residency rules. For Japanese banks, insurers, and government buyers, that means a US frontier model with a domestic integrator on the contract.

Read together, the three items map a sorting that accelerated this week. American export controls and now sanctions push Chinese AI vendors out of allied procurement. European governments respond by underwriting a continental champion they can certify as sovereign. Japan, with no domestic frontier lab of comparable scale, locks in an American partner through a national systems integrator.

Cohere, once a pure-play OpenAI rival, now has a German co-headquarters, a retail conglomerate as anchor investor, and two government endorsements. Aleph Alpha raised half a billion dollars in 2023 on promises of European independence. It now becomes a research arm rather than a standalone competitor. The deal consolidates Germany's most prominent foundation model effort into a Canadian-led structure.

The Anthropic-NEC tie-up arrives as Japan's regulators tighten cloud sovereignty guidance for public-sector AI procurement. NEC's existing government contracts give Anthropic a path through those rules that direct sales would not. The deal positions Claude as a credible default for Japanese ministries and large banks that need domestic-vendor sourcing.

02Sam Altman's Personal Apology to a Canadian Mining Town — and Anthropic's Pre-emptive Election Rules

Sam Altman sent a personal letter to residents of Tumbler Ridge, British Columbia, the mining town where a recent mass shooting took place. He told them he was "deeply sorry" that OpenAI "failed to alert law enforcement" about the suspect. The letter is the first time a frontier AI lab CEO has publicly apologized over a downstream user harm tied to the company's product.

Altman did not detail what the suspect's interactions with ChatGPT involved, or when OpenAI became aware of them. The letter focused on the gap between what the company knew and what it told police. He said OpenAI is reviewing internal escalation procedures.

The same week, Anthropic posted an update on its election safeguards. The post described policy refinements for how Claude handles election-related queries and content ahead of upcoming votes. Anthropic framed the change as a routine refresh of rules set in advance, not a response to any specific incident.

The two postures landed within days of each other. OpenAI described what it had failed to do after the fact. Anthropic published what it would not do, in advance. Neither company addressed the other's announcement.

Tumbler Ridge sits about 1,100 kilometers northeast of Vancouver and has a population of around 2,400. Local officials have not said publicly whether OpenAI contacted them before or after the shooting, or whether Altman's letter was the first formal communication. Canadian law does not currently require AI providers to report user threats to law enforcement.

The letter went out under Altman's name rather than from corporate communications. OpenAI has spent recent months pitching enterprise customers on safety guarantees including HIPAA compliance for hospital deployments. The Tumbler Ridge apology acknowledges a specific operational failure rather than reaffirming policy. The company has not said whether the incident will trigger reporting changes in other jurisdictions.

03The Murati–Meta Talent Trade Has Stopped Being Two-Way

Mira Murati left OpenAI as CTO to build Thinking Machines Lab. Since then, Meta has hired from her team and she has hired from theirs. Until this week, the trade ran both directions, engineers moving each way, neither side gaining ground.

That balance broke on Wednesday. Meta chief people officer Janelle Gale sent a memo, first reported by Bloomberg, telling staff that roughly 10 percent of the company will be cut in May. The number lands near 8,000 people. Gale's note also said Meta is closing around 6,000 open roles.

TechCrunch reported the same week that Meta's loss is now Thinking Machines' gain. Engineers who had been weighing the two offers are no longer weighing equally. A Meta offer now comes with a clock attached: the May review, the closed reqs, the chance your team gets folded into another.

Murati's recruiting pitch did not need to change. The lab she founded after walking out of OpenAI has spent 2026 hiring across post-training and infrastructure. What changed is what sits on the other side of the decision.

The flow inside Meta is concentrated. Generative AI org engineers, the ones working on Llama post-training and the consumer assistant, are the most mobile, according to TechCrunch. These are the same teams Meta paid nine-figure packages to assemble last summer. Some of those packages now belong to people interviewing at Murati's lab.

Meta has not said which orgs absorb the May cuts. Gale's memo described the goal as flattening management layers, not narrowing product scope. But the people closing the most expensive open reqs are the ones being asked to retain the most expensive existing staff.

That retention math gets harder every week Thinking Machines keeps shipping offers.