A new bipartisan advocacy group, Americans for Responsible Innovation, just launched to push a pragmatic framework for AI governance. Instead of reinventing the wheel, the group wants regulators to adapt proven playbooks from nuclear energy, cybersecurity, and banking—such as creating an auditing oversight board and restricting cloud training access.
Microsoft just cited the humble VCR in a motion to dismiss key claims in The New York Times’ copyright lawsuit. The company argued that large language models are like video recorders or personal computers—technologies with substantial lawful uses that shouldn't be outlawed due to potential misuse. Why it matters: If accepted, this classic defense could protect the entire generative AI sector from crushing copyright liability.
OpenAI just published a sharp rebuttal to Elon Musk's lawsuit, releasing old emails to push back against his claims. The records show Musk once pushed to merge OpenAI into Tesla, demanded controlling equity, and explicitly agreed that building AGI didn't require open-sourcing the science. Why it matters: OpenAI is dismantling Musk's narrative before the court, showing he backed their pivot until he walked away.
California regulators just gave Waymo permission to expand its driverless robotaxi service to highways in Los Angeles and the Bay Area at speeds up to 65 mph. The approval brushes aside objections from local transport agencies over recent safety scares, though Waymo says it has no immediate plans to send cars onto the highway right away.
Elon Musk just filed a lawsuit against OpenAI and Sam Altman, claiming the startup breached its founding agreement by chasing profits instead of benefiting humanity. Musk alleges OpenAI has become a closed-source subsidiary of Microsoft and asks the court to declare GPT-4 an AGI to block commercial licensing.
A Chinese internet court just found an AI service liable for copyright infringement after generating images substantially similar to Ultraman, ordering 10,000 yuan in damages. The ruling, seen as a world first, puts AI training data and output liability front and center for developers everywhere.
OpenAI just asked a federal judge to dismiss key parts of The New York Times' copyright lawsuit, arguing the publisher used "deceptive prompts" and exploited bugs to force ChatGPT into spitting out verbatim text. The Times fired back, claiming it was merely hunting for evidence of unlicensed copying.
Automattic is reportedly finalizing deals with OpenAI and Midjourney to feed user posts from Tumblr and WordPress.com into their AI models. An internal scrape reportedly scooped up a decade of posts, including some private content by mistake, prompting Automattic to promise a quick opt-out toggle. Why it matters: Platforms are desperate to cash in on user data, even at the cost of alienating the creative communities that built them.
Google's Demis Hassabis confirmed the company plans to restore Gemini's image generation of people in the next few weeks after pausing it over historical inaccuracies. Hassabis explained that well-intended diversity prompts were applied too bluntly across historical contexts, and fine-tuning is underway.
The Justice Department just appointed Princeton professor Jonathan Mayer as its first chief AI officer and chief science advisor. Mayer will build an expert team to help federal law enforcement keep pace with rapidly evolving tech. Why it matters: In line with the Biden administration's executive order, his team will help establish crucial fairness guidelines for AI in sentencing, parole, and surveillance.
Google just shut off image generation of people in its Gemini chatbot after the tool churned out historically inaccurate and overcorrected images. The company admitted its diversity safeguards went too far, causing the model to misapply diversity to specific historical contexts and become overly cautious.
Google DeepMind just created a new AI Safety and Alignment organization to tighten safeguards around Gemini and tackle future AGI risks. Led by UC Berkeley professor Anca Dragan, the U.S.-based unit will expand beyond London's existing alignment team to address current issues like misinformation and medical advice while researching superintelligent control. Why it matters: Google is scrambling to show regulators and enterprise buyers that its models can be trusted without flying off the rails.
Air Canada just lost a lawsuit after its AI customer service chatbot made up a fake refund policy for a grieving passenger. The airline tried to argue in court that the bot was a "separate legal entity" responsible for its own actions, but the tribunal flatly rejected it, ruling companies are fully liable for what their AI tells customers.
ElevenLabs just introduced new safeguards to prevent its AI voice tech from meddling in the 2024 elections. The company rolled out a "no-go voices" safeguard that blocks users from mimicking active political candidates, starting with US and UK races, and explicitly outlawed political robocalls and impersonation chatbots. Why it matters: With critical elections taking place worldwide, the pressure is on AI voice platforms to curb convincing deepfakes before they derail voter trust.
Twenty leading tech companies just teamed up to curb deceptive AI in elections. At the Munich Security Conference, companies including Meta, Google, Microsoft, and OpenAI pledged to coordinate on detecting and labeling misleading AI deepfakes targeting candidates and voters. Why it matters: With dozens of national elections underway, it's an unprecedented joint effort, though the voluntary accord notably avoids outright bans or binding penalties.
A federal judge just dismissed nearly all claims in Sarah Silverman and other authors' lawsuits against OpenAI, throwing out allegations of vicarious infringement and DMCA violations. The judge noted plaintiffs failed to show ChatGPT outputs were substantially similar to their works. OpenAI still faces direct infringement claims, but the authors' case just took a massive blow.
OpenAI just teamed up with Microsoft to terminate accounts tied to five state-affiliated threat groups. The hackers were caught using its models for basic tasks like debugging code, drafting phishing content, and researching open-source vulnerabilities. Why it matters: OpenAI found that AI currently offers only incremental advantages for cyberattacks over existing tools, but shutting them out early keeps the pressure on.
Mozilla just announced it is cutting around 60 jobs and shutting down side projects like 3D platform Hubs to refocus on its flagship browser. The organization is combining its Pocket, content, and AI/ML teams directly into Firefox to build 'trustworthy AI' features. Why it matters: After years of chasing diversified bets, Mozilla is retreating to its core base to survive the generative AI era.
Google just deployed a new machine learning algorithm to catch fraudulent reviews faster than ever. The system analyzes long-term patterns to flag sudden rating spikes and coordinated scam rings, helping block over 170 million policy-violating reviews in 2023—a 45% jump from the previous year.
OpenAI CEO Sam Altman warned that "subtle societal misalignments" — rather than rogue killer robots — are what keep him up at night. Speaking at the World Governments Summit, he reiterated calls for an IAEA-like oversight body while comparing today's AI to early black-and-white cellphones. Why it matters: The industry's top figure insists AI makers shouldn't be the ones writing the rulebook.
