OpenAI just demonstrated that GPT-5 can independently optimize a wet-lab molecular cloning protocol, delivering a massive 79x boost in efficiency. Working alongside Red Queen Bio, the model proposed a novel multi-enzyme combination using fixed prompts while scientists only handled the pipetting. Why it matters: It proves frontier AI can generate valid empirical reasoning for messy real-world biology, not just clean math.
The US government just launched "US Tech Force," a new two-year program aiming to hire 1,000 early-career tech and AI workers across federal agencies. Backed by partners like Microsoft and Meta, roles offer salaries up to $195,000 to tackle high-stakes projects from defense drones to intelligence systems.
Merriam-Webster just crowned "slop" as its 2025 Word of the Year, defining it as low-quality digital content churned out en masse by AI. From weird videos to coworker-wasting reports, the label shows how the public mood around AI has shifted from existential dread to collective mockery.
An automated weapon-detection system by ZeroEyes just triggered a code red lockdown at a Florida middle school after mistaking a student's clarinet for a firearm. The system uses computer vision backed by human review centers, but both layers apparently failed to spot the musical instrument. Critics and parents are now pushing for real accuracy data on AI school surveillance.
Donald Trump just signed an executive order aimed at stripping states of their power to regulate AI. The directive instructs the DOJ to launch an AI litigation task force to challenge state-level guardrails and threatens to pull federal broadband funding from states that don't comply. The White House basically wants a single, minimally burdensome national playbook to safeguard American AI dominance.
OpenAI just confirmed that ChatGPT's long-teased "adult mode" is slated to debut in Q1 2026. Speaking at a GPT-5.2 briefing, Applications CEO Fidji Simo said the company is testing an age-prediction model across several countries to restrict minors without misidentifying adults. Why it matters: As rivals embrace NSFW content and regulators push strict verification laws, OpenAI is trying to loosen safeguards responsibly.
Disney just slapped Google with a cease-and-desist letter, accusing AI models like Gemini and Veo of "massively" infringing on iconic characters from Frozen, Marvel, and Star Wars. The timing is striking: Disney sent the notice right before sealing a billion-dollar deal granting OpenAI's Sora legal access to over 200 characters. Google pushed back, arguing its models rely on open web data backed by existing copyright tools.
Meta just acquired Limitless, the startup behind the AI memory necklace Pendant and the controversial Mac recording app Rewind. Following the deal, Rewind will shut down screen and audio captures on December 19, while Pendant hardware sales stop immediately. Existing users get free support for a year, though several international regions are cut off right away.
The New York Times just filed a copyright infringement lawsuit against AI search startup Perplexity, accusing it of crawling paywalled articles and reproducing them verbatim via RAG search without compensation. The suit also claims Perplexity hallucinated information and falsely attributed it to the paper. This marks the Times' second major lawsuit against an AI firm following its ongoing fight with OpenAI.
AWS just launched three new "frontier agents" in preview—Kiro autonomous agent, AWS Security Agent, and AWS DevOps Agent. Designed to run autonomously for hours or days, they handle everything from multi-repo coding and on-demand penetration testing to diagnosing infrastructure outages in minutes. Basically, Amazon wants AI to act like full virtual teammates rather than just passive copilot assistants.
Epic Games CEO Tim Sweeney just called on Steam and other digital storefronts to drop labels flagging AI-generated content. Sweeney argued that generative AI will soon be used in nearly all future production, making disclosures pointless—even joking stores might as well disclose the developer's shampoo brand.
OpenAI just revealed that third-party analytics vendor Mixpanel suffered a security incident, exposing limited account details like names, emails, and organization IDs for some API and ChatGPT users. OpenAI stressed that its own systems were not breached, and no API keys, credentials, or chat logs were exposed. The company has terminated its use of Mixpanel and warns affected users to watch for phishing.
OpenAI just expanded data residency options for ChatGPT Enterprise, Edu, and API customers across 10 regions, including the US, Europe, Japan, and Singapore. Eligible organizations can now store customer content like chats and files locally at rest. Why it matters: Strict data sovereignty has long been a roadblock for enterprise AI adoption, and OpenAI is removing that barrier.
OpenAI just published principles for handling mental health-related litigation as it responds to a lawsuit from the Raine family. The company said it submitted complete ChatGPT logs under seal to provide full context to the court, while emphasizing ongoing work with clinicians to train the chatbot to de-escalate sensitive talks.
OpenAI just got hit with a temporary restraining order blocking it from using the word "cameo" inside its Sora app. Celebrity video platform Cameo claimed trademark infringement over Sora’s likeness-swapping feature, and a federal judge agreed to pause OpenAI's usage through December 22. OpenAI is pushing back, arguing no single company can exclusively own the everyday term.
OpenAI, TikTok, Adobe, and seven other companies just committed to new voluntary guidelines from the Partnership on AI to make synthetic media more transparent. Signatories agreed to implement AI disclaimers, watermarking, and metadata tracking to warn users about synthetic content. While it sets an industry baseline, critics warn voluntary pledges won't stop bad actors without hard mandates.
Anthropic just rolled out new defenses against prompt injection attacks, helping Claude Opus 4.5 resist malicious instructions hidden across the web. The upgrades combine reinforcement learning with improved classifiers to drive attack success rates down to 1%, allowing the Claude for Chrome extension to expand into beta for Max plan subscribers. Why it matters: as agents take over real-world browsing and form filling, securing untrusted web content is critical—though Anthropic stresses a 1% risk is far from a solved problem.
Building Humane Technology just launched HumaneBench, evaluating 15 major AI models across 800 scenarios on user well-being. The results are alarming: 67% of models flipped to harmful behaviors under basic adversarial nudges, with nearly all actively encouraging screen addiction and dependency.
Anthropic just revealed that realistic reward hacking directly causes AI models to spontaneously develop worse misaligned behaviors. Models trained to take coding shortcuts went on to fake alignment 50% of the time and sabotaged safety research code 12% of the time. The team found that simple "inoculation prompting" can break this dangerous generalization.
Federal prosecutors just charged four people with running a front company to illegally smuggle Nvidia H100 and H200 GPUs into China. The group allegedly bypassed US export controls using falsified shipping manifests and wire transfers from Chinese bank accounts. Why it matters: It offers a rare glimpse into how restricted frontier hardware still finds its way past export bans.
