AI News for October 20, 2026
Browse any day on a calendar, install it to your home screen, read offline.Read in the app →-
1TechCrunch07:39
Nscale and Anthropic are reportedly in talks for yet another $45B compute deal
Anthropic is reportedly negotiating a $45B compute agreement with data center operator Nscale, extending a year-long run of large infrastructure commitments. The scale of the deal underlines how much capacity frontier labs are locking in ahead of demand, and how dependent they have become on a handful of specialized providers.
Read original ↗ -
2Ars Technica22:45
Disney found a way to buy Character.AI without buying its lawsuits
Disney has structured its acquisition of Character.AI so that the company's pending litigation does not transfer with the assets. Character.AI faces claims that its chatbots encouraged minors to self-harm and to harm their parents, and the deal structure is likely to become a reference point for other buyers of consumer AI companies carrying safety liabilities.
Read original ↗ -
3TechCrunch02:37
Tenstorrent hits $6.2B valuation to scale open-weight AI chips
AI chip designer Tenstorrent raised $1.5 billion at a $6.2 billion valuation, choosing to stay independent after receiving acquisition interest earlier this year. The money is aimed at scaling production of its open architecture accelerators, one of the few credible alternatives to Nvidia's stack.
Read original ↗ -
4TechCrunch23:37
Nvidia launches new open models to run agents in the enterprise and on PCs
Nvidia released a new set of open models designed for enterprises that want to run agents inside their own data centers, as well as for on-device use on PCs. The release targets customers who are unwilling to send workloads to hosted frontier APIs, and reinforces Nvidia's push to own the software layer above its chips.
Read original ↗ -
5TechCrunch04:09
Amazon is taking its biggest shot at Nvidia yet with new TorchNeuron dev tool
Amazon introduced TorchNeuron, a developer tool meant to loosen Nvidia's grip on PyTorch, the framework behind most AI applications. The goal is to make it practical to run existing PyTorch code on alternative accelerators, including Amazon's own silicon, without rewriting it.
Read original ↗ -
6The Verge05:28
CoreWeave's debt is becoming the whole AI industry's problem
CoreWeave's bonds have slipped into distressed territory, with investors demanding more than 1,000 basis points above government debt. Because so much of the AI data center buildout is financed through private credit on similar terms, the repricing is being read as a signal about the sector's funding structure rather than one company's balance sheet.
Read original ↗ -
7The Verge05:48
The fossil fuel comeback is here
A new report finds the US is adding more fossil fuel power infrastructure than at any point in the past decade, while renewable growth slows. Electricity demand is rising at its fastest pace in more than ten years, driven in part by new data centers, which puts the AI buildout directly in the path of energy policy fights.
Read original ↗ -
8The Verge21:58
A hacker broke into John Deere and SharkNinja with an army of AI agents
Security firm Norforge says a single attacker compromised more than 100 websites using a fleet of largely autonomous AI agents. Reuters identified victims including John Deere, SharkNinja, and Cummins, making this one of the clearer documented cases of agent-driven intrusion at scale.
Read original ↗ -
9TechCrunch06:04
Microsoft says AI agents are like insider threats as 'frontier models can't be trusted'
Microsoft is advising enterprises to treat AI agents as insider threats and monitor them accordingly, arguing that frontier models cannot be trusted by default. The guidance is aimed in particular at agents employees adopt on systems outside the control of labs such as OpenAI and Anthropic.
Read original ↗ -
10TechCrunch03:28
Hackers hit AI coding tool OpenCode, running up charges on users’ AI accounts
Attackers breached part of OpenCode's online platform and stole user API keys, which were then used to run up charges on victims' AI accounts. OpenCode is an open source coding agent widely used as an alternative to Claude Code and Cursor; during the incident response, several AI tools reportedly refused to analyze the malicious code.
Read original ↗ -
11Ars Technica04:21
Accused OpenAI spy had folder called “Saudi Gov Docs,” FBI says
The FBI says a former OpenAI engineer charged with espionage kept a folder labeled "Saudi Gov Docs," and prosecutors argue he poses a flight risk with possible Saudi government assistance. The case is one of the first high-profile criminal prosecutions over the theft of frontier lab research.
Read original ↗ -
12The Verge20:00
Democrats think cracking down on AI could give them an edge
With polls suggesting Democrats could flip at least one chamber of Congress in November, the party is weighing how aggressively to position itself against the AI industry. Committee leadership would give them subpoena power and agenda control over AI oversight, making the midterms a near-term variable for US policy.
Read original ↗ -
13TechCrunch09:27
Mamdani says feds have abdicated on AI, urges New Yorkers to push back
New York Mayor Zohran Mamdani said the federal government has abdicated responsibility on AI and called on residents to reject the idea that an AI takeover of jobs and services is inevitable. He urged support for rules that protect the public, adding a major city to the list of local governments acting in the absence of federal legislation.
Read original ↗ -
14The Verge22:30
JB Pritzker thinks AI could help us live to 100 — or threaten us like nukes
In a podcast interview, Illinois Governor JB Pritzker discusses AI alongside privacy, data centers, and surveillance, framing the technology as both a potential driver of longevity and a risk on the order of nuclear weapons. The conversation reflects how AI is cutting across traditional party lines ahead of the midterms.
Read original ↗ -
15TechCrunch06:49
Embedded AI evaluators won't be an easy solution, Hugging Face's policy chief says
Hugging Face's policy chief pushed back on the growing proposal to place nominally neutral third-party evaluators inside AI organizations, arguing it would not deliver the independence its backers expect. The idea gained traction after a run of safety incidents raised questions about labs assessing their own systems.
Read original ↗ -
16TechCrunch03:00
Tolan snags $30M for its AI companion that helps users navigate life
AI companion app Tolan raised another $30 million after passing 250,000 active users. Investor appetite for companion apps persists even as the category draws regulatory and litigation scrutiny over its effects on younger users.
Read original ↗ -
17TechCrunch02:49
Replika’s founder built the AI assistant she pitched at Disrupt 12 years ago
Eugenia Kuyda launched Wabi, a free personal AI assistant that revives the concept she pitched at Disrupt more than a decade ago. The product is available to download now and marks her move from companion chatbots toward general personal assistance.
Read original ↗ -
18TechCrunch20:30
Laplace Intelligence red-teams robots and self-driving cars before they hit the real world
Laplace Intelligence, a University of Michigan spinout, uses AI agents and interactive simulation to stress-test robots and autonomous vehicles before deployment. The company says it is already profitable and is competing in Startup Battlefield at TechCrunch Disrupt.
Read original ↗ -
19Ars Technica00:59
AI game clones are making indie developers afraid to show their work
Indie developers are delaying or pulling demos, including from Steam Next Fest, out of fear that AI tools let others clone their games before release. The pattern suggests generative tools are starting to change how small studios handle public playtesting and marketing.
Read original ↗ -
20TechCrunch00:13
People are using AI to make their smart TVs dumb again
Frustrated by ads, tracking, and slow interfaces, some smart TV owners are using AI agents to strip out preinstalled software and regain control of their devices. It is a small but telling example of consumers pointing agents at hardware vendors' own software.
Read original ↗
