Top Stories
Federal Judge Blocks Pentagon’s Anthropic Ban, Cites ‘First Amendment Retaliation’
A federal judge granted Anthropic’s request for a preliminary injunction Thursday, blocking the Trump administration from designating the AI company as a national security risk.
U.S. District Judge Rita Lin issued a 43-page ruling that didn’t mince words. “Punishing Anthropic for bringing public scrutiny to the government’s contracting position is classic illegal First Amendment retaliation,” Lin wrote. She called the Pentagon’s rationale “Orwellian,” stating that “nothing in the governing statute supports the notion that an American company may be branded a potential adversary and saboteur of the U.S. for expressing disagreement with the government.”
The dispute traces back to a $200 million contract Anthropic signed with the Pentagon last July. When the company refused to allow Claude for fully autonomous weapons or mass surveillance of Americans, the administration responded by labeling it a “supply chain risk.” Court filings revealed the Pentagon told Anthropic the two sides were “nearly aligned” just a week before Trump publicly severed the relationship.
Lin delayed implementation of her ruling for one week to allow the government to appeal. The case sets important precedent for whether the executive branch can weaponize national security designations against AI companies that refuse certain military applications.
OpenAI Abandons Erotic Chatbot and Sora in Strategic Refocus
OpenAI is shelving its planned “adult mode” for ChatGPT just days after announcing the shutdown of its Sora video generator.
The erotic chatbot project, first floated by Sam Altman in October, would have relaxed content filters for verified adult users. An internal advisory council raised concerns about emotional overreliance on AI, harmful content scenarios described as “sexy suicide coach” situations, and inadequate safeguards for child safety and mental health.
The decision follows last week’s Wall Street Journal report that OpenAI is engaging in a “major strategy shift” to focus on business users and coders rather than consumer features. The Sora shutdown, attributed to unsustainable inference costs per generated minute, cost the company a prospective $1 billion partnership with Disney.
Sam Altman has stepped back from safety oversight to focus on fundraising and infrastructure — “building datacenters at unprecedented scale.” The company is expected to release a new model codenamed “Spud” in coming weeks.
Source: TechCrunch, Decrypt
Y Combinator Demo Day Showcases 190 Startups, 60% AI-Focused
Nearly 190 companies presented at Y Combinator’s Winter ‘26 Demo Day on Tuesday, with AI dominating the batch at roughly 60% — up from 40% in 2024.
The cohort reflects a notable shift: founders are increasingly building products designed not just for human users but for AI agents. Notable companies include ARC Prize Foundation, which creates benchmarks to measure progress toward AGI (already used by OpenAI, Anthropic, and Google), and Asimov, which collects human movement data to train humanoid robots.
The batch “tilts hard toward physical-world problems” including robotics, energy, agriculture, aerospace, and construction. One startup, Button, has built a tiny computer that connects to apps like email, Slack, and Salesforce and operates them via voice command.
Source: TechCrunch
Quick Hits
-
Apple-Google Siri deal takes shape: Apple is paying about $1 billion annually to use Google’s Gemini, with iOS 26.4 shipping a rebuilt Siri featuring complex reasoning and multi-step planning. Apple can distill Gemini into smaller on-device models.
-
Meta cuts continue: Bloomberg reports Meta is cutting several hundred more jobs across sales, recruiting, and Reality Labs. The company estimates $162-169 billion in 2026 AI infrastructure spending.
-
2026 layoffs climb: Over 59,000 tech jobs cut in 2026 so far, with more than 12,000 explicitly attributed to AI and automation. Oracle is evaluating 20,000-30,000 layoffs to fund AI infrastructure commitments.
-
90% don’t trust AI with data: A Malwarebytes survey found 88% of respondents won’t freely share personal information with AI tools like ChatGPT and Gemini.
-
MCP hits 97M installs: The Model Context Protocol crossed 97 million installs in March, signaling its transition from experimental standard to foundational agentic infrastructure.
Worth Watching
The government has one week to appeal Judge Lin’s Anthropic ruling. An appeal to the Ninth Circuit could escalate this into a landmark case on executive power over AI companies.
OpenAI’s strategic pivot away from consumer features — dropping Sora and adult mode to focus on enterprise tools — suggests the company sees its future in business productivity rather than creative applications. The “Spud” model release will reveal how that strategy translates to product.
Apple’s rebuilt Siri with Gemini underpinnings ships in iOS 26.4. This marks the biggest upgrade to Siri since launch and could finally make Apple’s assistant competitive with ChatGPT and Claude. The full rollout is expected with iOS 27 at WWDC.
The privacy trust gap (90% don’t trust AI with their data) represents a real obstacle for AI adoption. Companies that can demonstrate genuine data protection — like Apple’s Private Cloud Compute approach — may have a significant advantage.