Skip to content
Daily Brief News Daily Brief

6 AI stories from June 16, 2026: OpenAI burned $34B in 2025, Anthropic talks stall in DC, WaPo on planned export controls, Meta CTO admits atrocious reorg, GitHub turns to AWS, and DOJ backs xAI on national security grounds

· by Pondero Newsdesk · 6 stories

AI news daily brief: 2026-06-16

Six stories today: OpenAI's audited 2025 burn rate lands weeks before the S-1, Anthropic's DC talks end without a deal, the Washington Post traces the shutdown to a premeditated escalation, Meta's CTO admits the Applied AI rollout was atrocious, GitHub adds AWS capacity to absorb AI coding demand, and the DOJ frames Grok as a national security asset to block an environmental lawsuit.

OpenAI burned $34 billion in 2025 and posted a $38.5 billion net loss as its S-1 approaches

Journalist Ed Zitron published audited OpenAI financial documents on June 15, with the Financial Times independently verifying the figures, per Zitron's report at Where's Your Ed At. OpenAI booked $13.07 billion in revenue in 2025 against $34 billion in total costs, producing a $20.92 billion operating loss. The net loss attributable to the company came to $38.53 billion, driven partly by a $41.55 billion fair-value charge tied to the company's nonprofit-to-for-profit conversion. That is roughly 7.6 times the $5.09 billion net loss recorded in 2024. Research and development spending alone reached $19.18 billion, exceeding the company's full-year revenue. OpenAI paid Microsoft $17.2 billion across cost of revenue and R&D line items during the year. SoftBank paid OpenAI $867 million, Microsoft paid the company $303 million. A public S-1 is expected in late July or August at a target valuation above $1 trillion. The 42 state attorneys general who opened an investigation into OpenAI earlier in June now have audited financials to work with.

Full story: OpenAI 2025 audited financials

Anthropic's senior leaders met Trump officials June 15 and reached no deal on the Fable 5 shutdown

Co-founder and chief compute officer Tom Brown and policy chief Sarah Heck led Anthropic's side of talks with Trump administration officials on June 15, 2026, per The Information. No resolution was reached. Fable 5 and Mythos 5 remained offline, entering a fifth day of suspension. The Decoder's June 15 report noted that sources described the two sides as speaking in different languages, citing Axios reporting. Government officials have said Anthropic released Fable 5 without going through a Commerce Department clearinghouse that Trump's recent cyber executive order designated as the frontier model approval mechanism. Further sessions involving the CIA and science adviser Michael Kratsios were planned. Both sides said publicly they want a quick resolution. The June 12 Commerce Department directive barred Fable 5 distribution to any foreign national, leaving Anthropic no choice but to suspend access globally. Claude Opus 4.8 and other Claude models were not affected.

Full story: Anthropic talks June 15, no resolution

Washington Post: Trump officials planned Anthropic export controls for weeks after a China-linked firm requested Mythos access

A four-byline Washington Post investigation published late June 15 found that Trump administration officials began weighing sanctions on Anthropic weeks before the June 12 shutdown order, per the primary Washington Post report. The triggering event was a request to access the Mythos model from a firm the White House suspected had ties to China. That incident shattered what administration officials described to reporters as already-fragile trust in the company, and internal deliberations about sanctions began well before June 12. A companion Post piece reported that three days after Fable 5 launched publicly, the administration called Anthropic with a takedown demand, giving the company 90 minutes to comply and providing little explanation. The reporting establishes the shutdown was not an emergency response to a newly found jailbreak but the endpoint of a weeks-long deterioration in the government's relationship with Anthropic.

Full story: WaPo on planned Anthropic export controls

Meta CTO Andrew Bosworth called the Applied AI reorg atrocious in an internal memo and promised engineers stability

Meta CTO Andrew Bosworth circulated an internal memo acknowledging the company did an "atrocious" job rolling out its Applied AI division, per Wired reporting on June 15. The roughly 6,500-person unit, formed in March, drafted engineers through surprise emails with no meaningful ability to opt out. Those who resisted were told the alternative was resignation. Bosworth promised to cap manager spans at roughly 20 direct reports, reduce forced reassignments, give drafted engineers the ability to apply for other internal roles, and restore office perks including stacked snack kitchens and assigned desks. He also announced a company-wide hackathon in July. His memo followed an earlier one from Mark Zuckerberg acknowledging the changes caused distress and pledging no further company-wide layoffs in 2026. The structural promises Bosworth put in writing are now measurable commitments.

Full story: Bosworth memo on Applied AI reorg

Microsoft turned to AWS to stabilize GitHub after AI coding agents drove nine outages in May

Microsoft added Amazon Web Services capacity to GitHub to handle demand surges driven by AI coding tools and autonomous agents, per Business Insider on June 16. GitHub COO Kyle Daigle said in April the platform was processing 275 million commits per week, a pace that would put the full-year count at 14 billion versus 1 billion for all of 2025. The Register reported nine service incidents at GitHub in May 2026 alone, with further outages recorded in June. GitHub is mid-migration from its monolith to Azure, with 40 percent of monolith traffic on Azure as of May. The company is not abandoning that Azure migration, but the pace of AI-driven growth required an immediate capacity bridge through AWS. Whether GitHub formalizes a multi-cloud strategy or accelerates the Azure migration timeline is the primary question to watch.

Full story: GitHub AWS capacity

DOJ asked a court to dismiss the NAACP xAI turbine lawsuit by arguing Grok is vital to US military operations

The Justice Department filed to dismiss an NAACP Clean Air Act suit against xAI in a Memphis-area data center case, arguing xAI is integral to US military operations including the Iran War, per the Earthjustice press release on the NAACP's emergency motion. The NAACP, represented by Earthjustice and the Southern Environmental Law Center, sued xAI in April 2026 over natural-gas turbines operating without air permits at its Southaven, Mississippi site. The DOJ wrote that "it is the policy of the United States to sustain and enhance America's global AI dominance in order to promote human flourishing, economic competitiveness, and national security." The filing is the clearest statement yet of the federal government treating a frontier AI company as defense-critical infrastructure. Whether the court accepts the national security argument will determine whether the NAACP's pending preliminary injunction request, and the underlying environmental claims, proceed to a merits phase.

Full story: DOJ, xAI, and the NAACP lawsuit

Sources