Top AI Stories · November 15, 2026

Project Zero shortens bug deadlines; Tesla opens Cybercab orders

Google's Project Zero cuts its bug-disclosure deadline to 30 days by 2028, citing AI attackers. Tesla opens personal Cybercab orders with a required subscription. Plus 5 more stories.

Listen to this brief

Free preview · first 0:30
0:00 / 0:30

Audio & video are paid features

Plus unlocks audio streaming and PDF downloads. Pro adds offline MP3 downloads, video, certificates, and more.

Plus adds:
  • Audio streaming
  • Downloadable PDFs
  • All AI Playbooks
  • Personalized content
Pro also adds:
  • Certificates of completion
  • Audio MP3 downloads
  • Video lessonssoon
  • & More…soon

Google's Project Zero says AI has made its old bug-disclosure timeline unsafe, and Tesla's first personal Cybercab orders come with a mandatory monthly subscription. Federal officials also cleared the FAA's new AI routing tool of any role in the November 13 ground stop.

  1. 1

    Google's Project Zero cuts its bug-disclosure deadline to 30 days by 2028, citing AI attackers

    Google's Project Zero security team is retiring the policy that gave vendors 90 days to fix a reported vulnerability plus 30 more days for the patch to roll out. Under the new rule, details go public a fixed number of days after the report, whether or not a fix exists: 90 days now, 60 days from February 2, 2027, and 30 days from January 3, 2028. The team blames AI. Language models let attackers find the same bugs and work out the flaw from a patch almost immediately, so the extra rollout time no longer keeps anything secret. AI tools have also raised the team's own output to "several vulnerabilities a week in targets where we once found a few vulnerabilities a year." Grace periods are gone. The only exceptions are bugs already exploited in the wild, bugs others have already published, and early release at the vendor's request.

  2. 2

    Tesla opens personal Cybercab orders, with a mandatory $299 monthly subscription

    Tesla has begun inviting select customers to order a Cybercab for personal use, The Verge reported from order-page screenshots posted by Tesla's Robotaxi account. The two-seater costs $29,990 plus a required Autonomy Subscription of $299 a month, and it is sold only in cities where Tesla already runs its Robotaxi service. Buyers pay a $1,000 non-refundable deposit for a December delivery window that is not guaranteed. The terms bar commercial use, so owners cannot put the car on Tesla's ride-hailing network to earn money, which was a central promise when Elon Musk unveiled the car in 2024. Because the Cybercab has no steering wheel or pedals, cancelling the subscription leaves it unable to drive on public roads. The Verge puts five years of ownership at roughly $47,600 before charging and insurance.

  3. 3

    A router, not the FAA's new AI routing tool, caused the nationwide ground stop

    Federal officials say a contractor's network upgrade, not artificial intelligence, caused the ground stop that halted US departures for more than four hours on November 13. The Homeland Security secretary had said agencies were examining whether SMART, the AI routing tool the FAA had just expanded nationwide, played a role. On November 14, Transportation Secretary Sean Duffy said a router failed where L3Harris was replacing old copper lines with fiber, flooding the network with about 1,000 connection attempts every four to five seconds and disrupting communications at 17 of the FAA's 22 en route centers. FAA Administrator Bryan Bedford said there was "zero evidence of anything AI-related." More than 9,000 US flights were cancelled that day and more than 5,000 the next.

  4. 4

    A petition to bring back Jensen Huang passes 115,000 signatures, including Elon Musk's

    More than 115,000 people had signed a petition by the afternoon of November 14 asking Nvidia's board to reinstate Jensen Huang while an independent investigation runs, TechCrunch reported. The petition site counts 5,965 current Nvidia employees, 1,108 former employees and 63,063 shareholders among them, and names Elon Musk, Epic Games CEO Tim Sweeney, Replit CEO Amjad Masad and Y Combinator CEO Garry Tan as supporters. It says Huang "was removed without meaningful due process." Abroad, Singapore's Ministry of Home Affairs said it had investigated the Megaspeed executive at the center of the case over shipments of servers to Malaysia, found too little evidence of an offense, warned her for giving inaccurate information about her role, and barred her from returning. China's foreign ministry has called the spying allegations groundless.

  5. 5

    A judge lets a lawsuit proceed claiming memory makers squeezed DRAM supply for AI chips

    A federal judge in California refused on November 14 to dismiss a proposed class action alleging that Samsung, SK Hynix and Micron conspired to cut supply of ordinary computer memory while shifting factory capacity to high-bandwidth memory (HBM) for AI accelerators. Judge Noël Wise found that long-term HBM supply deals, refusals of long-term contracts with conventional buyers, Micron's closure of its Crucial consumer brand, and the failure to expand output despite record prices together plausibly point to an agreement. She rejected booming AI demand as an obvious innocent explanation at this stage. The complaint says conventional memory prices rose 171.8 percent in a year. The allegations are unproven, and the companies must answer by December 7.

  6. 6

    Thinking Machines delivers the Inkling 2 Small weights it promised, under Apache 2.0

    Mira Murati's Thinking Machines Lab has published the weights for Inkling 2 Small on Hugging Face, a week after launching the model on its Tinker platform with closed weights and a promise to release them once outside safety testing was done. The weights arrived on November 12 under a standard Apache 2.0 license, with no revenue, user or country limits. The model card also links the lab's acceptable-use policy, which says anyone downloading the weights agrees to follow it. Inkling 2 Small is a mixture-of-experts (MoE) model with 594 billion total parameters, 17 billion active per token. The recommended checkpoint needs about 500 GB of GPU memory, such as two Nvidia B300s. Together, Fireworks, Baseten, Cloudflare and others already host it.

  7. 7

    Helen Toner: AI safety cannot depend on trusting any one CEO

    In an interview with The Verge's Nilay Patel, Helen Toner, who runs Georgetown's Center for Security and Emerging Technology and was one of the OpenAI board members who voted to fire Sam Altman in 2023, said she would make the same choice again. "If our hopes for the future are pinned on any one individual being trustworthy, then we're pretty screwed," she said, arguing for structures that involve many more people. She also said the heavy spending against midterm candidates who talked about AI regulation largely failed, because voters care about the issue and it has become bipartisan. The conversation was recorded before Nvidia opened its investigation into Jensen Huang.

Get Top AI Stories by email

The day's most important AI news — free, daily, unsubscribe anytime.

Share
Spot a typo or have feedback?Share feedback

Sources

  1. 1.What forced the US to ground all planes? A router switch, we now know — Al Jazeera · November 15, 2026
  2. 2.thinkingmachines/Inkling-2-Small — Hugging Face · November 12, 2026
  3. 3.FAA investigating whether new AI technology contributed to nationwide ground stop — News4JAX · November 13, 2026
  4. 4.Asian governments react to Nvidia CEO Jensen Huang's exit — TechCrunch · November 14, 2026
  5. 5.Judge Refuses to Dismiss DRAM Supply Conspiracy Suit Against Samsung, SK Hynix, Micron — LitNews · November 14, 2026
  6. 6.Policy and Disclosure: 2026 Edition — Google Project Zero · November 13, 2026
  7. 7.Tesla is inviting some people to buy a Cybercab 'for personal use' — but read the fine print — The Verge · November 14, 2026
  8. 8.Elon Musk, other tech execs sign petition calling on Nvidia to reinstate Jensen Huang — TechCrunch · November 14, 2026
  9. 9.The OpenAI board member who fired Sam Altman says we still can't trust AI CEOs — The Verge · November 14, 2026
  10. 10.Inkling 2 Small — Thinking Machines Lab · November 5, 2026

AI disclosure: Researched and drafted with AI; reviewed and edited by the AI Pro Playbook editorial team before publishing. Sources above link to original publishers.

🧭Recommended for you