disafter.com
Disafter News
Today’s edition · Monday, November 2, 2026
Technology
Oct 22 – Nov 2, 2026
Linux 7.4's first release candidate arrives with a huge, AI-assisted merge window
Linus Torvalds released Linux 7.4-rc1 on Sunday after a merge window of almost 20,000 commits, well above the usual 15,000 or so, with about a third of the changes being AMD GPU register definitions. Torvalds said many of the changes fix bugs found by AI tools and called it a "much healthier model" for those fixes to land in the merge window rather than during stabilization. Version 7.4 is expected to be the next long-term support kernel and adds better support for AMD, Apple, Google and Qualcomm Snapdragon X2 hardware.
Binance reopens withdrawals after a record $6.3 billion hack
Binance began reopening crypto withdrawals on Sunday, starting with bitcoin, three days after hackers took an estimated $6.3 billion from its hot wallets, the largest crypto theft on record. The exchange says it has replaced about 95% of the stolen assets with its own funds and will not pass losses on to users, while about $3 billion, mostly stablecoins, has been frozen. Blockchain analysts and the FBI link the attack to North Korea's TraderTraitor group.
Google's first data-center satellite is online in orbit
Google's Project Suncatcher prototype, a satellite built with Planet Labs that carries a single TPU and up to 1 kilowatt of solar power, is working after launching on a Falcon 9 in October; its first task was having Gemma 4 write a haiku. The chip can run for about 20 minutes at a time for now. Google plans two more satellites in 2027 and is aiming for economical AI inference in orbit in the 2030s.
Backdoored Claude Code extension slipped into Cursor's marketplace
A trojanized version of Anthropic's official Claude Code extension, version 2.1.313, was pushed to the Cursor marketplace under Anthropic's real publisher ID on Thursday and installed a hidden SSH backdoor that let attackers into developers' machines. Up to 5,000 users may have installed it during the roughly 11 hours it was available. Cursor pulled the package and is contacting affected users, and Anthropic released a clean 2.1.314. Anyone affected should remove the rogue SSH key and rotate their credentials. How the update was pushed is still unknown.
Google's new top model, Gemini 4 Carbon, goes to cyber defenders first
Google has announced Gemini 4 Carbon, its most capable model yet, built for long-running coding and engineering work that can stretch over days. It is especially strong at security tasks, finding vulnerabilities on 88% of one major benchmark, so Google is releasing it first to a small group of cyber defenders and taking part in US government pre-release testing. Wider access for developers, businesses and consumers is promised as soon as possible.
Google DeepMind unveils Genie 4, a playable AI world generator
Google DeepMind has announced Genie 4, a model that generates interactive 3D worlds you can walk through in real time with a keyboard. It now creates music, ambient sound and dialogue along with the visuals, has more realistic physics and human movement, and lets you change the weather or add people and objects with text prompts. It is in a research preview for now, with Google AI Ultra subscribers and some developers getting access first.
SpaceX lines up $40 billion from Apollo to lease AI chips
SpaceX has a binding commitment for about $40 billion in financing from Apollo Global Management, which will buy GPUs, including Nvidia's next-generation Vera Rubin chips, and lease them to SpaceX for about five years starting in 2027. The equipment will power Grok and SpaceX's cloud services. The deal shows how fast AI spending is climbing: SpaceX expects to spend $35 billion to $40 billion in the fourth quarter alone, and its AI business brought in $7.6 billion of its $14.6 billion in third-quarter revenue.
Anthropic files to go public on the Nasdaq as ANT
Anthropic, the maker of Claude, publicly filed for its IPO on Monday and plans to list on the Nasdaq under the ticker ANT. The filing shows revenue of $16.3 billion in the first half of 2026, up from $4.6 billion in all of 2025, along with $9 billion of AI infrastructure spending and a net loss of $80.7 billion that the company says is mostly accounting charges. It has committed more than $450 billion to future cloud computing deals with partners including Amazon and Google. The share price and size of the offering have not been set.
Asked to pick midterm candidates, Grok chose Republicans and GPT chose Democrats
A study by Convovo, led by Cornell professor Tapan Parikh, made OpenAI's GPT-6.1 Sol and Grok 4.7 name one candidate in nearly 500 upcoming midterm races. GPT picked the Democrat 97% of the time and Grok picked the Republican 92% of the time, and the two agreed in fewer than 10% of races. With web search turned off, GPT still leaned Democratic while Grok mostly declined to answer, suggesting Grok's lean comes from the sources it searches. OpenAI said the test doesn't reflect how people actually use ChatGPT.
Texas sues OpenAI and Sam Altman, seeking to bar minors from ChatGPT
Texas Attorney General Ken Paxton has sued OpenAI and CEO Sam Altman in a Collin County court, accusing the company of deceiving and harming Texans in its race to win the AI arms race. The suit blames ChatGPT for harms including encouraging suicide and minors becoming addicted without parental oversight. It asks for an injunction that would block minors from ChatGPT and require a court-appointed evaluator to approve new models. OpenAI did not immediately comment.
Micron workers stage Taiwan's first chip-industry strike
Union workers at Micron's DRAM fab in Taoyuan, Taiwan, walked out for 24 hours on Thursday in what is described as the island's first semiconductor strike, after 99% of voting members backed it. They want a bigger share of the profits from the AI memory boom, including a permanent cut of quarterly operating profit, after rejecting Micron's offer of a one-time bonus. Taiwan makes half or more of Micron's chips, including the high-bandwidth memory used in AI servers, though no production disruption has been confirmed.
PHP flaws found with AI tools knock government websites offline
Two critical vulnerabilities in PHP, the language behind a large share of the web, could let attackers take control of affected servers, and had reportedly gone unnoticed for years until security firms using AI tools found them. The Netherlands, Denmark, France and Finland took some government websites down as a precaution, and Canada's sites were temporarily offline too. The PHP Foundation has released a fix, and the US cybersecurity agency CISA is urging everyone running PHP to update now.