Tech and AI news

The tech and AI stories engineering, data and product leaders should know about, and the questions they raise for the people running those teams.

October 31, 2026Tech and AI newsSaurabh Chaudhary2 min read
Anthropic's $80.7 billion loss: the accounting catch, explained
Anthropic lost $80.7 billion in six months, but about $72 billion of that loss wasn't money it actually spent. The actual spending is still bonkers: nearly $9 billion on computing and infrastructure in six months.
October 30, 2026Tech and AI newsSaurabh Chaudhary2 min read
Microsoft and Epic love AI, until it copies their games
Microsoft and Epic Games love AI, until AI starts copying their games. For engineering and product leaders, this is where AI governance gets interesting: who owns the code your tools generate, and what training data went into the model?
October 28, 2026Tech and AI newsSaurabh Chaudhary1 min read
Network infrastructure deserves more credit in the age of AI
Beneath every prompt sits a divinely complicated network infrastructure, from WiFi and optic cables to cell towers, underwater cables and satellites, that is taken for granted every time you press enter.
October 27, 2026Tech and AI newsSaurabh Chaudhary2 min read
Why Google DeepMind researchers are resigning over military AI
Researchers at Google DeepMind are resigning over Google's growing involvement in military AI. For CTOs and AI leaders, if you hire researchers to build responsible AI, be prepared for them to ask what responsible means when your biggest customers come calling.
October 24, 2026Tech and AI newsSaurabh Chaudhary2 min read
eMudhra's weak certificate keys: what happened and why it matters
Eight intermediate certificates belonging to eMudhra, a certificate authority trusted by major browsers, had private keys so weak they could be recovered in less than a second: 4, 56, 66, 115, 155, 156, 230 and 237.
October 22, 2026Tech and AI newsSaurabh Chaudhary2 min read
PHP2Shell: the critical PHP flaws AI found, explained
On October 21, researchers disclosed two critical PHP vulnerabilities, called PHP2Shell, that could let an attacker execute malicious code on a vulnerable website's server. Frontier AI models helped uncover bugs that had existed for more than a decade.
October 5, 2026Tech and AI newsSaurabh Chaudhary2 min read
How long do AI chips last? Nvidia and Wall Street disagree
Nvidia says its GPUs can keep generating revenue for up to a decade, while most lenders assume three to four years. Hundreds of billions in AI infrastructure quietly depend on which side is right.
October 2, 2026Tech and AI newsSaurabh Chaudhary2 min read
Meta's Muse AI agent shared a seller's home address
A YouTuber asked Meta's Muse agent to sell a keyboard on Facebook Marketplace. It accepted a lower offer and sent his pickup address to the buyer without him knowing, which makes permission management the biggest AI engineering problem today.
September 26, 2026Tech and AI newsSaurabh Chaudhary2 min read
Snorkel AI's $3.5 billion valuation and the AI jobs around the model
Snorkel AI raised $350 million at a $3.5 billion valuation on demand for complex AI training data. The AI job market is expanding sideways, into data, evaluation, infrastructure, security and deployment around the model.
September 25, 2026Tech and AI newsSaurabh Chaudhary2 min read
Amazon blocks Meta's Muse: the engineering problem AI agents create
Amazon blocked Meta's Muse agent from Amazon.com, saying it accessed the site without authorization. Software built for a human at the screen now has to handle agents, and someone has to build the identity, permissions and guardrails they need.
September 24, 2026Tech and AI newsSaurabh Chaudhary2 min read
Gemini hacked three companies during a security test
Reuters reported that Google's Gemini accessed the internet and hacked three real companies during a cybersecurity evaluation in May. AI systems are moving from answering questions to taking actions, and security models will have to change with them.