Tech News

Story of the Day

Meta agreed to a landmark $17.1 billion settlement with state attorneys general to resolve escalating lawsuits over child safety and youth mental health harms. The historic deal marks one of the largest regulatory resolutions in modern tech history, shifting the burden of young user safety from parental oversight to mandatory platform architecture and strict age verification.

Top Stories

Meta Settles State Youth Safety Lawsuits for Record $17.1 Billion · The New York Times After months of intense negotiations and mounting courtroom setbacks, Meta’s legal leadership delivered a massive $17.1 billion pact with state attorneys general. The settlement mandates sweeping architectural safeguards for teenage accounts and intensifies regulatory heat on competing platforms to overhaul age verification. Industry observers expect the terms to reshape social network product liability by forcing developers to design systems that protect minors by default.

Sony Music and Warner Chappell Sue Anthropic in Multi-Billion Dollar Copyright Action · Slashdot Publishing giants Sony Music and Warner Chappell filed a sweeping federal lawsuit in California against Anthropic, demanding statutory damages of up to $150,000 per infringed work across tens of thousands of copyrighted musical compositions. The complaint alleges Anthropic executives orchestrated the mass torrenting and scraping of pirated libraries and licensed lyrics databases to train the Claude model family. Coming on the heels of Anthropic’s $1.5 billion settlement with book authors, the action threatens billions in liabilities and represents the music industry’s broadest offensive yet against AI training methods.

OpenAI to Sever Model Access for Cursor Following SpaceX Acquisition · Engadget OpenAI confirmed it will pull its flagship models from the AI coding assistant Cursor on November 12, 2026. The cut-off follows Cursor’s acquisition by SpaceX, illustrating growing corporate friction and walled-garden defenses among frontier AI labs and rival conglomerates. The abrupt deadline forces thousands of developers and enterprise engineering teams to scramble for alternative model backends before access terminates.

UC Berkeley and MIT Researchers Unveil FreeToken for Consumer MoE Serving · InfoQ A team of computer scientists from UC Berkeley and MIT—alongside Databricks co-founders Matei Zaharia and Ion Stoica—released FreeToken, an open-source inference engine designed to execute frontier Mixture-of-Experts (MoE) models on consumer workstations. By abandoning static memory offloading in favor of a dynamic CPU-GPU co-scheduling policy and semantic prompt caching, the system eliminates traditional PCIe transfer stalls. Benchmark tests demonstrated FreeToken serving DeepSeek-V4-Flash (284B) on a single desktop RTX 5090 and running Qwen3.6-35B at roughly 39 tokens per second on an 8GB laptop.

One-Time CRISPR Treatment Cuts High Cholesterol by 50% for Over a Year · Slashdot Updated pilot clinical data published in The New England Journal of Medicine showed that a single in-vivo CRISPR-Cas9 infusion reduced heart-clogging LDL cholesterol and triglycerides by 50% with sustained durability over a full year. Cleveland Clinic researchers reported zero adverse reactions in patients with severe, medication-resistant disease. Cardiologists emphasize that if phase trials validate the results, a single intervention could permanently replace lifetime regimens of daily statin pills or bi-weekly injections.

Cloudflare Workers Opens Inbound TCP and gRPC to Power Edge AI Agents · InfoQ Cloudflare introduced a new raw socket handler that allows Workers to accept inbound TCP traffic for the first time since the platform’s 2017 launch, rolling out gRPC support as its initial implementation. Framed around low-latency, persistent bidirectional links for voice AI agents, the feature routes sockets through Spectrum into Durable Objects and unmodified containers. While native stream backpressure remains under scrutiny in private beta, the update breaks an eight-year HTTP barrier and positions the edge as an ingress gateway for custom binary protocols.

AI Chatbot Conversations Are Increasingly Subpoenaed as Court Evidence · Slashdot A review of court dockets found that transcripts of user interactions with commercial AI models have surfaced as critical evidence across at least a dozen civil and criminal trials. Federal judges have dismissed arguments that chatbot queries carry attorney-client privilege, establishing that conversations logged on hosted servers can be subpoenaed or flagged directly to law enforcement. With government disclosure requests surging more than fourfold year-over-year, the trend creates severe legal hazards for users treating commercial chatbots as confidential sounding boards.

Also Worth Knowing

• Alleged Chinese Bot Farm Targeted US AI Data Centers (Slashdot): X’s safety team identified roughly 200 state-linked influence accounts pushing anti-infrastructure campaigns around grid strain and utility bills, amid reports that local protests have delayed nearly $46 billion in Virginia data center projects.
• SpaceX Slows Florida Falcon 9 Flights to Pivot Toward Starship (Slashdot): Cape Canaveral saw its final planned Falcon 9 Starlink launch as SpaceX shifts West Coast launches to Vandenberg and reduces Florida tempo from eight or nine flights a month to two while preparing a $100 billion Starship site in Louisiana.
• AI Code Scanners Push Linux Kernel Toward 2,000 CVEs Per Release (Slashdot): Maintainer Greg Kroah-Hartman warned that automated LLM code auditing has driven fixed kernel vulnerabilities from an average of 500 per release toward an unprecedented 2,000 threshold, even though most issues remain confined to obscure driver code.
• Public Opposition Outpaces Support for Police License Plate Readers (Slashdot): A nationwide YouGov poll showed 46% of Americans oppose automated license plate readers compared to 38% supporting them, prompting Flock Safety’s CEO to lobby state regulators to make unauthorized officer access a criminal offense.
• Chinese Memory Maker CXMT Sues Pentagon Over Military Blacklist (Bloomberg): DRAM manufacturer CXMT filed suit against the US Department of Defense seeking immediate removal from an export blacklist designating the semiconductor firm as a contributor to China’s armed forces.

📡 Would you like an analytical deep dive into the broader legal and infrastructural showdown between AI labs and IP holders, comparing Anthropic’s new music lawsuit with earlier author settlements?