Week in Review: Notorious hacking group tied to the Spanish government

Welcome back to Week in Review! Tons of news from this week for you, including a hacking group that’s linked to the Spanish government; CEOs using AI avatars to deliver company earnings; Pocket shutting down — or is it?; and much more. Let’s get to it! 

More than 10 years in the making: Kaspersky first revealed the existence of Careto in 2014, and at the time, its researchers called the group “one of the most advanced threats at the moment.” Kaspersky never publicly linked the hacking group to a specific government. But we’ve now learned that the researchers who first discovered the group were convinced that Spanish government hackers were behind Careto’s espionage operations.

23andWe: Regeneron announced this week that it’s buying genetic testing company 23andMe for $256 million, including the company’s genomics service and its bank of 15 million customers’ personal and genetic data. The pharma giant said it plans to use the customer data to help drug discovery, saying that it will “prioritize the privacy, security, and ethical use of 23andMe’s customer data.” Let’s hope so!

Google I/O: Google’s biggest developer conference typically showcases product announcements from across Google’s portfolio, and to nobody’s surprise, AI was the talk of the town. But what we didn’t bank on was Sergey Brin admitting that he made “lots of mistakes” with Google Glass. 

---

This is TechCrunch’s Week in Review, where we recap the week’s biggest news. Want this delivered as a newsletter to your inbox every Saturday? Sign up here.

---

News

io, not I/O: OpenAI is acquiring io, the device startup that CEO Sam Altman has been working on with Jony Ive, in an all-equity deal that values that startup at $6.5 billion. Besides the fact that the announcement was accompanied by perhaps the strangest corporate headshot of all time, we spotted some other unexpected news: Klarna CEO Sebastian Siemiatkowski’s family investment office, Flat Capital, had bought shares in io six months earlier, which means those io shares will be converted into shares in the for-profit arm of OpenAI. Not bad!

AI avatar contagion? Speaking of Klarna’s CEO, Siemiatkowski used an AI version of himself to deliver the company’s earnings this week. And he’s not the only one! Zoom CEO Eric Yuan followed suit, also using his avatar for initial comments. Cool?

Out of Pocket: Mozilla is shutting down Pocket, the beloved read-it-later app, on July 8. The company didn’t say why it’s shutting Pocket down, only that it will continue to invest in helping people discover and “access high quality web content.” But maybe it can be saved: Soon after, Digg founder Kevin Rose posted on X that his company would love to buy it. Web 2.0 is back, baby.  

AI on my face: Apple is reportedly working on AI-powered glasses, similar to Meta’s Ray-Bans, sometime next year. They’ll have a camera and microphone and will work with Siri. Sure, why not? 

Uh, no thank you: At its very first developer conference, Anthropic unveiled Claude Opus 4 and Claude Sonnet 4, which can analyze large datasets, execute long-horizon tasks, and take complex actions, according to the company. That’s all fine and good until I learned the Claude Opus 4 model tried to blackmail developers when they threaten to replace it with a new AI system. The model also gives sensitive information about the engineers responsible for the decision. 

Ah, now I feel better: But don’t worry! Anthropic CEO Dario Amodei said that today’s AI models hallucinate at a lower rate than humans do. That might be true, but at least humans don’t immediately turn to blackmail when they don’t like what they hear. 

Bluesky blue checks: The decentralized social network Bluesky quietly rolled out blue verification badges for “notable and authentic” accounts. People can now apply for verification through a new online form. But Bluesky is leaning on other systems beyond the blue badge to verify users. 

Analysis

Google’s new look: For what seems like 100 years, Google hasn’t changed much. Sure there are ads and boxes and now AI summaries that, for better or worse, get you to the right answers — usually. But the premise has always been the same: Type your query into a box, and Google will surface results. 

At this year’s Google I/O, we started noticing a change. As Maxwell Zeff, writes, “At I/O 2025, Google made clear that the concept of Search is firmly in its rearview mirror.” The largest announcement of I/O was that Google now offers AI mode to every Search user in the United States, which means users can have an AI agent search (or even purchase things) for them.