The Voss Report — June 15, 2026

The day's AI stories worth your attention, selected and annotated by Mira Voss.

[AI researchers launch Sequent because "alignment is not on track"](https://jack-clark.net/2026/06/15/import-ai-461-alignment-is-not-on-track-frontiercode-and-synthetic-research-interns/) Import AI Former UKASI and Timaeus researchers are forming a new nonprofit, Sequent, targeting $100–150M to run a diversified portfolio of alignment bets — scalable oversight, learning theory, heuristic arguments, game theory — on the explicit premise that frontier lab alignment programs are "essentially reactive" and will not yield principled confidence before ASI arrives.

[The US government is quietly letting a key data center regulation expire](https://www.wired.com/story/the-us-government-is-letting-a-key-data-center-regulation-expire/) Wired The Federal Data Center Enhancement Act sunsetting in September with no replacement means there will be no federal standards governing how the government's own data centers operate — at the precise moment AI infrastructure is absorbing a historically unusual share of federal attention and capital.

[Salesforce acquires Fin (formerly Intercom) for $3.6 billion](https://www.salesforce.com/news/press-releases/2026/06/15/salesforce-signs-definitive-agreement-to-acquire-fin/) Salesforce Fin's Apex model reportedly outperforms GPT-5.4 and Claude Sonnet 4.6 on customer support resolution benchmarks, which tells you something about what purpose-built vertical models can do — and explains why Benioff moved at this price rather than building.

[Anthropic's Claude is now available through Apple's Foundation Models framework](https://platform.claude.com/docs/en/cli-sdks-libraries/libraries/apple-foundation-models) Anthropic / Apple A new Swift package lets developers route calls to Claude through the same LanguageModelSession API that drives Apple's on-device model — Apple is not in the request path, requests bill to Anthropic accounts directly — which makes every iOS and macOS app a potential frontier-model interface with no platform intermediary.

[Skydio CEO: Silicon Valley shouldn't draw red lines for autonomous drone use](https://www.theverge.com/podcast/949195/skydio-ceo-adam-bry-autonmous-drones-china-red-lines-military) The Verge Adam Bry's argument — that unilateral ethical restraint by US drone makers creates an asymmetry China won't reciprocate — is the same logic the defense AI industry uses to resist every proposed constraint; what's notable is how comfortably it's now said aloud by a tech CEO rather than a general.

[Britain announces social media ban for under-16s starting 2027](https://www.nytimes.com/2026/06/15/world/europe/social-media-bans-worldwide.html) The New York Times The UK joins Australia and a handful of other governments in treating age-based platform exclusion as a governance tool — a significant shift in how democracies are willing to override platform autonomy, with implications for AI-native social products that don't yet exist as a regulatory category.

The Voss Report runs daily. For original reporting, see The Signal, The Mirror, and The Becoming.