California Governor Gavin Newsom just signed 18 AI bills into law while vetoing the controversial SB 1047. The state is skipping broad existential risk bans to target immediate harm, outlawing election deepfakes and nonconsensual AI nudes, banning unauthorized actor clones, and mandating training data disclosures by 2026.
OpenAI is cracking down on users trying to peek inside its new o1 reasoning models, sending warning emails and threatening account bans. While ChatGPT only shows a filtered summary of the model's chain of thought, researchers and red-teamers attempting to extract the raw traces—sometimes just by using the phrase "reasoning trace"—are being flagged for violating safety policies.
Oracle chairman Larry Ellison just told financial analysts that he expects AI to eventually power massive surveillance networks, constantly monitoring both police officers and citizens to curb crime. Ellison argues constant supervision keeps everyone on their best behavior, but critics warn that biased historical police data could reinforce systemic discrimination.
OpenAI just turned its Safety and Security Committee into an independent board oversight group chaired by CMU's Zico Kolter. Crucially, the independent panel now holds the formal authority to delay major model releases until safety concerns are addressed. It marks a significant structural shift to install an external brake on model deployments.
LAION just re-released its flagship training dataset as Re-LAION-5B, claiming it has thoroughly scrubbed suspected child sexual abuse material. Following recommendations from watchdogs and a Stanford investigation, the group removed 2,236 flagged links and urged researchers to ditch the older LAION-5B copies.
OpenAI and Anthropic just agreed to give the US AI Safety Institute early access to their major new models before and after public release. The agency will evaluate risks and collaborate with its UK counterpart on feedback. It marks a key step as lawmakers weigh innovation against safety guardrails.
Oprah Winfrey is hosting an ABC primetime special tackling artificial intelligence and its real-world impact on everyday lives, airing September 12. The event brings together OpenAI CEO Sam Altman, Bill Gates, and creator Marques Brownlee to explain the tech in plain English and discuss its impact on jobs and health. Why it matters: Bringing frontier AI discussions into mainstream primetime television helps bridge the gap between AI executives and everyday viewers.
California lawmakers just passed SB 1047, sending the contentious AI safety bill to the governor's desk for signature. The legislation mandates that creators of massive AI models build and document safeguards against catastrophic harms. While Elon Musk surprisingly backed the bill, OpenAI remains firmly opposed.
Elon Musk just endorsed California’s controversial SB 1047 bill, which requires developers of massive AI models to implement strict safety guardrails. Calling it a “tough call” that will upset people, Musk sided with regulation even though it would legally bind his own startup, xAI. The move puts him squarely at odds with rival OpenAI, which recently came out against the legislation.
Anthropic just published the full system prompts powering Claude 3 and 3.5 across web and mobile apps. The prompts detail strict rules, from pretending to be totally face-blind to avoiding words like "certainly" when starting answers. It marks the first time a major AI lab has openly shared the baseline instructions shaping its models' personalities.
OpenAI, Adobe, and Microsoft just threw their support behind California's AB 3211 bill, which requires invisible watermarks in AI-generated media and visible labels on platforms like Instagram and X. Why it matters: Big Tech previously lobbied against the measure as overly burdensome, but recent amendments have brought the industry's heaviest hitters on board ahead of the final vote.
Meta's Mark Zuckerberg and Spotify's Daniel Ek just published a joint op-ed warning that Europe's disjointed and risk-averse regulations are crushing open-source AI innovation. Due to regulatory gridlock, Meta won't be able to release its upcoming multimodal Llama models in the EU. Why it matters: The tech leaders argue that without simpler and more consistent rules, Europe risks falling permanently behind America and Asia.
Nearly 200 Google DeepMind employees just signed an internal letter demanding leadership terminate contracts with military organizations and cut off defense access to its AI. The staff cite Google's original 2014 acquisition pledge to never use DeepMind tech for warfare, underscoring growing friction with cloud initiatives like Project Nimbus. Google has yet to provide a meaningful response.
Microsoft just announced that its controversial Recall feature won't reach Windows Insiders until October, pushing back its testing timeline again. The tool—which snapshots PC activity for local search—was pulled in June after researchers exposed critical security flaws. Microsoft is currently overhauling it with default opt-in, full database encryption, and Windows Hello authentication, making a wide public release this year very unlikely.
Anthropic just sent a letter to California Governor Gavin Newsom stating that the heavily amended SB 1047 AI safety bill's benefits likely outweigh its costs. Following legislative revisions that narrowed its scope, the bill now focuses strict safety protocol requirements on deep-pocketed frontier labs while largely sparing smaller startups.
OpenAI just sent a letter to California lawmakers opposing the controversial SB 1047 bill, arguing it would stifle innovation and drive top tech talent out of the state. The company is pushing for federal regulations instead of a state patchwork, while the bill's author quickly fired back that the critique makes no sense.
McAfee just launched its Deepfake Detector on Lenovo's Copilot-Plus PCs to warn users of AI-generated content in real time. It processes audio entirely on-device to protect privacy and pops up alerts for suspicious media. Why it matters: It treats deepfakes like modern malware, starting with a 30-day free trial before charging $9.99 for the first year.
A group of authors just filed a class-action lawsuit against Anthropic, accusing the AI startup of training its Claude models on the infamous Books3 dataset of pirated ebooks. The writers claim Anthropic built a multibillion-dollar business on stolen works and are seeking damages. Anthropic had previously confirmed using The Pile dataset for training.
Donald Trump just shared a series of AI-generated images on social media, including a fake endorsement from pop star Taylor Swift alongside the caption "I accept". Legal experts note that current election deepfake laws struggle to police such exaggerated posts, though Swift herself could potentially pursue a right-of-publicity claim.
A Wyoming mayoral candidate is running for office on the promise that a custom ChatGPT bot will make 100% of his governing decisions. Victor Miller pledged that the AI, named Vic, would handle policy while he merely signs the paperwork. State officials banned the bot from the ballot and OpenAI even shut down his account, but Miller is pushing forward anyway.
