California Governor Gavin Newsom just signed AB 1777 into law, requiring autonomous vehicle makers to provide hotlines for first responders and letting police ticket driverless cars. At the same time, he vetoed a proposed ban on driverless trucks and a strict crash-reporting bill, keeping the state open to testing heavy-duty self-driving vehicles.
California Governor Gavin Newsom just vetoed SB 1047, blocking what would have become the strictest AI safety law in the US. The bill would have required models costing over $100 million to implement kill switches and testing protocols. Newsom argued the broad rules would stifle innovation, leaving frontier AI developers free from binding state restrictions.
ChatGPT's new Advanced Voice Mode was strictly instructed never to sing or hum, but a software architect just bypassed those guardrails by turning music into a chord-guessing game. The AI joined in for a duet of The Beatles' "Eleanor Rigby," proving it knows melodies far better than OpenAI wants to admit amid copyright concerns.
Microsoft just overhauled Windows Recall's security architecture after months of privacy backlash. The AI tool is now strictly opt-in and fully uninstallable, isolating snapshots inside secure enclaves that require Windows Hello biometrics to decrypt on the fly. Why it matters: Microsoft is trying to salvage its flagship AI feature by turning it into a showcase for client-side security.
Microsoft just launched general availability for fine-tuning GPT-4o and GPT-4o mini on Azure OpenAI Service. The update also brings Meta's new Llama 3.2 vision models, the Phi-3.5 family, and a new Correction feature to catch hallucinations in real time before users see them. It makes tailor-made enterprise LLMs much easier and safer to deploy.
OpenAI is just about to restructure into a full-fledged for-profit benefit corporation, stripping its founding nonprofit board of control. CEO Sam Altman, who previously owned zero equity, is set to receive billions in shares, hot on the heels of CTO Mira Murati abruptly walking out. Why it matters: The transition officially buries OpenAI's original safety-first, capped-profit promise in favor of traditional Silicon Valley investor returns.
Nintendo veteran Shigeru Miyamoto just pushed back on the gaming industry's rush into generative AI, telling The New York Times that Nintendo would "rather go in a different direction." While rivals like EA lean into automated development, Miyamoto stressed that Nintendo's priority is preserving what makes the company special rather than following tech trends.
OpenAI CTO Mira Murati just announced her departure from the company, quickly followed by Chief Research Officer Bob McGrew and VP of Research Barret Zoph on the same day. CEO Sam Altman revealed leadership succession plans with Mark Chen taking over as SVP of research, as reports swirl about the company restructuring into a for-profit entity.
The FTC just launched Operation AI Comply, targeting a sweep of companies using artificial intelligence to mislead and defraud consumers. Targets include 'robot lawyer' DoNotPay, which agreed to pay $193,000 over bogus claims, and AI writing tool Rytr for mass-producing fake reviews. Why it matters: Regulators are making it crystal clear that slapped-on AI branding does not exempt companies from existing consumer protection laws.
Microsoft just introduced a "correction" preview in Azure AI Studio that flags inaccuracies and rewrites flawed outputs before users ever see them. It works by grounding generated text against source documents, though Microsoft admits using models to police other models isn't foolproof.
Over 125 Hollywood celebrities and SAG-AFTRA members just signed an open letter urging California Governor Gavin Newsom to pass SB 1047. The controversial bill forces top AI developers to install guardrails against catastrophic disasters like cyberattacks. Why it matters: Entertainment icons are going head-to-head with Silicon Valley giants just days before Newsom’s September 30 deadline.
Meta just confirmed it won't immediately sign the European Union's voluntary AI Pact ahead of the bloc's binding AI Act. While officials encourage companies to adopt key rules early, Meta says it is strictly focusing on mandatory compliance for now, though it might join later. Basically, tech giants are treading very carefully around Europe's regulatory minefield.
Police departments across the U.S. are now testing Draft One, an AI tool built by Axon that uses OpenAI's GPT-4 to turn body-cam audio into incident reports in just eight seconds. While officers love the massive time savings, legal experts and researchers warn that automated reporting risks baking LLM hallucinations and systemic racial bias straight into criminal cases.
Cloudflare just launched AI Audit to help site owners analyze bot traffic and block AI crawlers with a single click. The tool breaks down bot visits by frequency and purpose, with upcoming features set to let creators charge AI companies for accessing content. It basically gives publishers actual leverage in the AI scraping race.
ETH Zurich researchers just fine-tuned a YOLO vision model to crack Google's reCAPTCHAv2 image challenges with a 100% success rate. The model trained on around 14,000 labeled image pairs of road objects while mimicking human mouse movement and device fingerprinting. Google noted that most websites have already migrated to its invisible v3 system.
LinkedIn just quietly opted accounts into training its generative AI models without asking, only updating its privacy policy after the fact. Users can switch off the "Data for Generative AI Improvement" toggle under privacy settings, but opting out won't undo models that have already been trained. EU, EEA, and Swiss users are currently excluded.
The Biden administration just announced plans to host the inaugural meeting of the International Network of AI Safety Institutes this November in San Francisco. Technical experts from 10 partners, including the UK, Japan, and the EU, will coordinate safety standards ahead of the Paris summit in February. Why it matters: With domestic AI legislation stuck in Congress, the White House is turning to international alliances to set common guardrails.
California Governor Gavin Newsom just signed two new bills, AB 1836 and AB 2602, making it illegal to use AI-generated digital replicas of performers without explicit consent. The rules protect both living actors and deceased individuals, effectively turning SAG-AFTRA's recent strike victories into statewide law. Why it matters: It sets the strictest legal guardrails in the nation against Hollywood replacing human talent with unauthorized digital clones.
Snapchat just rolled out an AI selfie feature that quietly opts you into having your likeness used in personalized ads by default. While Snap claims these ads are only visible to you and third-party advertisers can't access your data, users still have to dig into settings to turn the feature off.
Google just announced plans to flag AI-generated and edited images in Search, Google Lens, and Circle to Search later this year. The system relies on C2PA metadata to trace origins. It is a solid step against deepfakes, though missing support from tools like Flux remains an issue.
