And just like that, half the Internet went down.

Amazon Web Services, the cloud computing arm of the retailing giant, experienced a massive outage in the wee hours of Monday that knocked out what felt like everything. 

Amazon things, of course—from Alexa to Ring camera services. OpenAI’s ChatGPT. Airtable. Burger King. Canva. Coinbase. Delta. Duolingo. Epic Games’ Fortnite. Lyft. McDonald’s. Monarch. Perplexity. Reddit. Robinhood. Signal. Snapchat. Taco Bell. Toast. Uber. Venmo. Zoom. 

Nevermind various bank, telecom, and government services inside and outside the U.S.  

As of this writing, most of the outage has been resolved, and many reliant services are back online. Others are still picking up the pieces.

Amazon said the outage occurred in its “US-EAST-1” AWS region, a.k.a. its longstanding data center in northern Virginia. As it so happens, it’s the same facility that suffered during its last outages of this scale, in 2021 and 2023.

The breadth of the outage certainly left folks wondering if it was really so good to have so much of the Internet in one company’s control. At about 30%, Amazon controls the single largest slice of the global cloud infrastructure pie.

“If a company can break the entire internet, they are too big,” wrote Sen. Elizabeth Warren on social media. “Period.” —AN

Three weeks after OpenAI released its Sora 2 video generation AI model, the company announced that it’s working with major actor unions and talent agencies to defend against deepfakes.

OpenAI says it has “strengthened guardrails around replication of voice and likeness” in the wake of the creation of several unauthorized clips imitating actor Bryan Cranston (known for playing Breaking Bad’s Walter White, among many others).

“OpenAI expressed regret for those unintentional generations,” the company said, adding: “All artists, performers, and individuals will have the right to determine how and whether they can be simulated.” 

(Pro tip: Don’t anger the guy famous for saying, “I am the danger.”)

On Sept. 30, OpenAI launched Sora 2 with an opt-in policy for the use of an individual’s voice and likeness. But its opt-out stance on studios’ intellectual property faced immediate blowback from an array of Hollywood organizations, including talent agencies. 

It reversed its policy on Oct. 3. Still, the deepfakes persisted. 

This latest agreement—which includes the United Talent Agency (UTA), Creative Artists Agency (CAA), and the Association of Talent Agents—aims to create some urgency around the issue. And perhaps gets a certain Heisenberg off its case. —AN

It seemed like a picture-perfect moment.

Getty Images and Shutterstock, two of the largest stock media services, announced in January that they’d combine in a $3.7 billion deal. 

It was a bid to bulk up as big tech companies’ investments in generative artificial intelligence promised to eviscerate their longstanding image, video, and music licensing business.

But the deal requires regulatory approval—and on Monday, the U.K. Competition and Markets Authority said it had concerns about the merger, forcing the companies to address them by Oct. 27.

Does a combined company substantially lessen competition in the U.K.? According to the CMA, Getty is the “clear market leader” and Shutterstock is “one of the few material alternatives.”

Moreover, “the CMA has not seen evidence that GenAI players are either currently, or likely to be in the next few years, an alternative to stock content for a significant proportion of demand.”

Could the companies divest assets to competitors PA Media/Alamy and Adobe to get the deal through? We’ll find out soon. —AN

—Anthropic launches Claude Life Sciences. AI tools and support for researchers.

—Brain-computer interface restores vision in some blind patients. Science Corp.’s PRIMA involves a retinal implant and special eyewear.

—DeepSeek releases OCR model, a.k.a. a “vision language” AI model for text in images.

—Reid Hoffman and David Sacks are fighting. A continued clash of the Silicon Valley titans in the wake of that essay by Anthropic’s Jack Clark.

—Google “Trusted Testers” are real. Fifteen superfans will walk around with disguised pre-release Pixel hardware.

—Ticketmaster will limit users to one account in the wake of an FTC lawsuit about scalpers.

—Is OpenAI “too big to fail”? On the art of the deal and “playing the egos of Silicon Valley’s giants off one another.”

Thoughts or suggestions? Email newsletters@fortune.com