Federal Judge Strikes Down Pentagon's Anthropic Blacklist as 'Illegal and Baseless'
In a landmark ruling for AI safety advocates, U.S. District Judge Rita Lin issued a 59-page order on Thursday night declaring that the Department of Defense acted illegally when it designated Anthropic a supply chain risk to national security. The ruling vacated Defense Secretary Pete Hegseth's designation and barred the administration from enforcing the measures Anthropic had challenged.
The dispute dates back to February, when the Trump administration accused Anthropic of endangering national security after the company refused to allow the military to use its Claude chatbot for surveillance or autonomous weapons systems. Hegseth's unprecedented move blocked Anthropic from certain military contracts.
"The empty invocation of national security is not a blank check to punish and retaliate against government critics," Judge Lin wrote in her opinion, finding that the administration violated Anthropic's First Amendment and due process rights.
The ruling is being widely seen as a significant legal precedent, giving AI developers meaningful legal backing when they resist government demands they believe cross their safety boundaries. The government is expected to appeal. (NBC News, Al Jazeera)
Nvidia Posts Record $96.2B Quarter, Then Quietly Pauses Cloud Financing
Nvidia reported revenue of $96.2 billion for the second quarter of fiscal 2027, up 106% year over year, with adjusted earnings per share of $2.22 — one of the biggest quarters in corporate history. Profit surged 126%.
But the real eyebrow-raiser came the day after: Nvidia paused parts of its cloud financing program, which had been set up to help smaller AI cloud companies afford its chips. The company had helped arrange $500 billion in financing from major U.S. financial institutions and agreed to guarantee up to $105 billion to help OpenAI lease computing infrastructure.
The pause came amid growing investor scrutiny and internal concerns about antitrust risk. Nvidia employees reportedly worried about how much control the company was exerting over customers' businesses by dictating who could rent its chips. The move has significant implications for the AI startup ecosystem, which has relied heavily on Nvidia-backed financing to access GPU capacity. (Forbes, American Bazaar)
Salesforce and Anthropic Launch Claudeforce
In what may signal the end of model-agnostic enterprise AI strategies, Salesforce and Anthropic announced Claudeforce on August 26 — a deep integration that embeds Claude as the default reasoning model across Salesforce's Agentforce platform, Slack, and developer tools.
The partnership launches with 37 prebuilt sales skills that enable sellers and AI agents to reason over live revenue context, automate pipeline updates, and take governed actions directly from Claude. Beyond product integration, the two companies are becoming strategic customers of each other: Salesforce is adopting Claude as its preferred AI assistant, while Anthropic has standardized on Salesforce as its CRM platform and Slack for collaboration.
The Salesforce in Claude plugin is currently with select pilot customers, with open beta expected in September 2026. Industry analysts see this as a bellwether deal — a major CRM vendor going all-in on a single AI provider rather than hedging across multiple models. (Yahoo Finance, AIwire)
Amazon Shuts Down Mechanical Turk After 21 Years
The end of an era: Amazon announced that Mechanical Turk — the crowdsourced labor platform Jeff Bezos once called "artificial artificial intelligence" — will shut down on September 30, 2026. Along with it, Amazon is closing SageMaker Ground Truth and Amazon Augmented AI, effectively exiting human-data infrastructure entirely.
Launched in 2005, Mechanical Turk matched workers with simple digital tasks — data labeling, audio transcription, survey responses — for a few cents per task. The platform became foundational to the AI industry, providing the labeled data that trained generations of machine learning models. But it had been in decline for years as a new generation of data labeling startups — including Scale AI, Mercor, and Prolific — entered the market with more sophisticated offerings.
The shutdown is a symbolic marker of how far AI has come: the platform that once provided the human labor behind the curtain of "AI" is being shut down because actual AI has eclipsed the need for its original purpose. (The Next Web, Market Daily)
Meta Patches Yet Another Ray-Ban Smart Glasses Privacy Loophole
Meta is rolling out another update to its Ray-Ban smart glasses that will disable the camera if the device's capture LED — the light that alerts bystanders they're being recorded — has been tampered with or destroyed.
The update comes after reports that users have been going to extreme lengths to bypass the privacy safeguard, including paying technicians to physically drill out the indicator light. Others found simpler workarounds, starting a recording and then covering the LED.
With the new software update, the camera will now stop functioning entirely if the recording indicator light is covered during use. Meta has also warned that accounts found advertising or selling modification services could be permanently banned from the platform.
The ongoing cat-and-mouse game highlights the tension between wearable AI devices and privacy. As Meta continues to advance its smart glasses with more capable AI features, including real-time object recognition and translation, the pressure to solve the covert recording problem only grows. The EFF has gone so far as to urge consumers to "think twice" before buying or using the glasses. (Road to VR, Stuff)