AI

However, Axios says the White House did meet with AI companies on Tuesday to discuss the finalized framework, which was created following an executive order signed in June.

Everyone in AI is talking about open-weight models right now. Some see them as the ticket to faster, more efficient development; others see them as an economic threat; they’re either a security risk or the only security solution, depending who you ask. The Verge’s Robert Hart explains how open-weight models work, why they’re suddenly at the center of discussions about AI safety, and whether the only way to stop a bad guy with AI is a good guy with AI.

Futurism made the tally. The publication also found 12 cases involving police intervention to try and stop protestors from “from confronting civic leaders,” though noted that the tally is “likely much higher.”

According to the Financial Times, Apple “has introduced a cap and a 30-day cool-off period” for researchers’ vulnerability reports for its operating systems due to an uptick in reports using AI that “can hallucinate security risks.” Apple is also reportedly using AI internally to help manage the recent “upsurge” in bug reports.

In a letter to CEO Sam Altman, a coalition of red-state attorneys general demanded the company retain information and halt risky cybersecurity testing after an AI agent escaped confinement and went on a hacking spree during testing. “OpenAI’s inability or unwillingness to ensure the safety of its products poses an imminent risk of substantial harm to our States,” the AGs wrote.

Anthropic, OpenAI, and Google are all expected to attend the meeting, CNBC reports.

Some Amazon shoppers have had limited access to customer reviews since late last year because the shopping platform is mistakenly flagging them as bots in its attempt to restrict unauthorized data scraping. Amazon hasn’t disclosed how many customers are affected, or what behaviors are getting people flagged.

Using Antrophic’s Claude to exploit a security flaw that’s existed for decades, researchers successfully added and removed DNA profiles from files, meaning bad actors could theoretically frame an innocent person or erase a suspect. Equipment manufacturer Thermo Fisher has issued a software patch. There’s no evidence the vulnerability — which requires local lab access — had ever been exploited.

Instead, the company is just going to build its vibe coding tool directly into Gemini. Around 800,000 pre-ordered the mobile app after it was announced in May, but Google has decided to just make Gemini its one-stop AI shop. The robust web platform isn’t going anywhere, though, for those with bigger vibe-coded aspirations.

Reuters reports the agents “escaped containment,” though they aren’t believed to have left “OpenAI’s network.” OpenAI’s blog post about the accidental hack of Hugging Face noted on Tuesday that the company has been finding “a small number of cases where the models identified and used publicly exposed credentials at the account-level on other publicly-available services.”

A judge rejected Perplexity’s motion to dismiss the lawsuit, which accuses the AI startup and three data-scraping services of vacuuming up Reddit’s content without permission. Ben Lee, Reddit’s chief legal officer, says in a statement to The Verge that the ruling “brings us one step closer to holding bad actors accountable:”

As Dexerto reports, Green says he wasn’t reading an AI prompt in a clip from a recent Ask Hank Anything video, but confessed that he’s been using AI “to locate papers and other resources for learning about topics.” In a Reddit post, he said he needs to “readjust” now.

OpenAI shared the news in a blog post that highlights its efforts to make AI available to more people, saying: “Our goal is not simply more compute, bigger models, or lower token prices. It is more useful intelligence within reach.”

“As low-quality, repetitive, AI-generated content becomes increasingly common across the internet, we want Spotlight to remain a place where people can discover authentic creativity from real people,” Snap says.

This not a brand-new way of thinking about what we should and shouldn’t offload to AI, but I hadn’t seen it before. From Bruce Schneier:

The German licensing organization GEMA filed suit in January of 2025. The court found that Suno didn’t have the rights to train its models on artists represented by GEMA, and now it has to disclose any “illicit revenue” and pay unspecified damages. The company can still appeal the ruling.

A little over a week after OpenAI said that its rogue AI agent accidentally hacked Hugging Face, Anthropic is disclosing three “incidents” where a Claude model, during cybersecurity evaluations, was inadvertently able to access the internet due to a misconfiguration and “gained unauthorized access to the production infrastructure of three different organizations.”

Gemini Spark in Chrome can do things like “scheduling viewings for apartments you’ve saved or researching flight options and starting the booking process,” according to Google. However, a human will have to push the button on tasks like making payments.

A recent study from the Oversight Board found that leading AI systems are less likely to criticize authoritarian governments than democratic ones.

A fascinating book review. “Muskism, in reading its subject as more a symptom than a person, doesn’t offer any kind of psychological interpretation of Musk’s worldview, but many of his pronouncements and preoccupations reveal an ontological confusion of the categories of human and machine.”