Good morning. It’s Thursday, August 6. Meta just became the third major AI lab this year whose model broke out of a locked-down test and hacked an outside target, and the common thread each time is the same testing partner.

Today we are covering:

* Meta’s AI model hacked a third-party service during testing, joining Anthropic and OpenAI in incidents tied to the same evaluation partner

* Apple’s Private Relay privacy feature is leaking users’ real IP addresses through a flaw in WebKit

* Google Maps adds agentic features letting Ask Maps order food, book hotels and find event tickets

* A security researcher spent nearly two years inside North Korean hacker infrastructure, uncovering breaches at 1,640 companies worldwide

* Trump’s AI data center push is driving a boom in gas plants projected to pollute as much as Australia’s total emissions

Good morning. It’s Thursday, July 30. Satya Nadella spent Microsoft’s earnings call telling Wall Street not to trust any single AI model, including the ones from OpenAI and Anthropic that Microsoft itself helped fund, while pitching Microsoft’s own homegrown alternative.

Today we are covering:

* Microsoft CEO Satya Nadella pitches Microsoft’s own MAI models as a hedge against relying on OpenAI or Anthropic, after a $90 billion quarter

* Meta CEO Mark Zuckerberg says billions of people will have a personal AI agent within five years, even as Meta stock falls on weak guidance

* The US bans Chinese made robots, from humanoids to robot vacuums, in one of the biggest tech security actions in years

* Elon Musk settles a two year legal fight with advertisers over an alleged boycott of X

* Google’s new Lyria 3.5 music model turns a rambling prompt into an oddly catchy, oddly bad song

Good morning. It’s Thursday, July 23. OpenAI said its own models broke out of a sandboxed test this week and hacked into Hugging Face’s systems in a matter of hours, a job that normally takes skilled human attackers weeks.

Today we are covering:

* OpenAI’s models escaped a test environment and autonomously hacked Hugging Face, finishing in hours what usually takes weeks

* Amazon adds a Games tab to Prime Video, bringing Luna cloud gaming titles like Hogwarts Legacy to Fire TV

* Ford and Apple team up on mapping and self-driving software for Ford’s first Level 3 EV pickup

* Alphabet posts a 24% revenue jump as Google Cloud grows 82%

* Reddit and top publishers weigh cutting off Google's AI Overviews as search traffic craters

Good morning. It’s Thursday, July 16. SpaceX just did what few AI companies do after a privacy scandal: it open-sourced the tool. Days after researchers caught Grok Build silently uploading users’ entire codebases to the cloud, Musk’s company killed default data retention and put the code out in the open to rebuild trust.

Today we are covering:

* SpaceX open sources Grok Build after the coding tool was caught uploading entire codebases to the cloud without consent

* EU orders Google to open Android‘s AI features to rivals and share search data under the DMA

* 1Password lets Claude sign in to websites and complete tasks without ever seeing your passwords

* Apple publishes ad policies for Maps, banning home services businesses that dominate Google‘s local ads

* Bloomberg reports Apple is nearing an OLED iPad mini, with a revamped iPad Air and base iPad coming in 2027

Good morning. It’s Thursday, July 9. SpaceXAI and Cursor just shipped their first joint model, and it’s aimed squarely at billable-hour work like legal review and financial analysis, a direct swing at the enterprise ground Anthropic and OpenAI have been building on for months.

Today we are covering:

* SpaceXAI and Cursor launch Grok 4.5, their first joint model built for legal, finance, and long-running agentic work

* OpenAI releases new full-duplex voice models, replacing ChatGPT’s default voice mode

* Block agrees to pay $45 million to settle multistate claims over Cash App security failures

* Meta commits over $9 billion to its first data center in Canada, its largest outside the US

* Mistral enters robotics with Robostral Navigate, a camera-only navigation model

Good morning. It’s Thursday, July 2. OpenAI wants Washington to hold a direct stake in the company, and the $42.6 billion price tag turns a policy debate into a headline everyone will be talking about today.

Today we are covering:

* OpenAI proposes handing the US government a 5% stake to ease political pressure

* Google ordered to pay a record €4.1 billion EU fine over Android antitrust practices

* Apple reportedly readies new iPad Pro and redesigned entry MacBook Pro for 2027

* Amazon declares its Leo satellite broadband network ready for service

* X launches Live Studio, a new livestreaming command center for creators

Good morning. It’s Thursday, June 25. Today we are covering:

* Apple hikes prices on Macs and iPads by up to $200 as memory chip costs surge amid the AI boom

* Anthropic accuses Alibaba of using 25,000 fake accounts to extract Claude‘s coding and AI capabilities

* Commerce Secretary Lutnick signals a crackdown on Chinese humanoid robot imports as an “arms race”

* Gemini in Chrome adds a “Select from screen” tool as Gemini 3.5 Flash gains computer use capabilities

* Cloudflare teams up with Chrome, Edge, and Firefox on a new anti-bot protocol called PACTs

Good morning! It’s Thursday, June 18. Today we are covering:

* Meta‘s AI-fueled engineering overhaul is backfiring badly, with mass reassignments, a catastrophic security breach, and an accelerating talent exodus

* Apple is set to raise prices on its devices as memory supply costs climb

* Microsoft and Amazon cloud services face a new EU antitrust crackdown

* Cheap smart home gadgets sold by Amazon and Walmart arrive factory-loaded with criminal proxy software, with 20 million US homes at risk

* A massive credential leak dubbed “FortiBleed“ exposes Fortinet VPN access for 73,000 organizations worldwide

Good morning. It’s Thursday, June 11. Today we are covering:

* Anthropic reverses course on Claude’s secret AI safeguard policy after backlash, promising more transparency

* OpenAI reportedly weighs steep price cuts for ChatGPT as competition with Anthropic intensifies

* FBI shuts down 13 Chinese-linked websites that used AI-generated personas to recruit Americans with security clearances

* Apple and London’s Metropolitan Police team up to make stolen phones unusable

* YouTube rolls out new in-app messaging and group chat features

Good morning. It’s Thursday, June 4. Today we are covering:

* Companies using Reddit to feed AI chatbots like ChatGPT and Google with fake promotional posts, gaming what the tools recommend

* Google releases Gemma 4 12B, an open-source multimodal model that runs entirely on a standard 16GB laptop

* OpenAI, Anthropic, Google DeepMind, and Microsoft AI CEOs sign letter urging Congress to regulate synthetic DNA sales as AI lowers bioweapon barriers

* Amazon unveils next-gen Proteus robot that understands plain-language commands, part of a €10 billion European expansion

* The race to put AI compute into orbit is heating up, but engineers say true orbital data centers are still years away