PHP Exploits Hit Public Services & AI Uncovers Surge of Vulnerabilities - Tech News (Oct 23, 2026)

A startup backed by some of Silicon Valley's biggest names is cloning dogs right now, with the first puppies due early next year, and they're meant for the battlefield. More on that in a moment. Welcome to The Automated Daily, tech news edition. The podcast created by generative AI. I'm TrendTeller, and today is October 23rd, 2026. We've got fresh fallout from the PHP security crisis, a rough day for AI stocks, a shake-up at OpenAI, SpaceX eyeing your phone bill, and an encouraging result in cancer research. Let's get into it.

PHP Exploits Hit Public Services

Let's start with security, and an update on the PHP vulnerability story. The damage has now spread into the real world in a very visible way. Banks, airlines, railways, tax authorities and other public services in several countries have taken websites offline or walled them off while they respond to active attacks. It's a clear reminder of what happens when a flaw lands in software that sits underneath a huge slice of the internet.

AI Uncovers Surge of Vulnerabilities

And it may not be a one-off. Security experts say artificial intelligence is helping researchers find far more vulnerabilities than in past years. That's useful in principle, but it also means a bigger flood of disclosures to patch, and more chances for outages and exploitation. Recent finds have touched Node.js, Plex, LiteLLM, Railway and certificate infrastructure, on top of PHP.

Claude Proves Code Mathematically Safe

There is a more hopeful side to AI and security. Anthropic, working with a company called Theorem, says it used Claude to produce machine-checked mathematical proofs that code in 123 open-source libraries can't crash under defined conditions. Formal verification like this has usually been too slow and costly to use widely. Along the way, Claude also turned up real bugs that years of testing had missed. The bigger idea is moving software safety from 'we tested it a lot' toward 'we can prove it.'

AI Spending Fears Sink Nasdaq

Over to the markets, where the worries about AI spending we've been tracking came to a head. The Nasdaq fell 2.6 percent, its steepest one-day drop since early June, with chipmakers and AI-linked names leading the way down. Meta was a big part of the mood. Its shares slid 7 percent after it announced plans to issue non-voting shares to raise money for AI, and it reported its first cash outflow in more than a decade. Add rising borrowing costs for the data center buildout, and investors are asking harder questions about whether all this spending will pay off. Tesla, oddly, surprised with positive free cash flow and trimmed its 2026 spending plans, yet its stock still dipped about 3 percent.

Meta Non-Voting Shares, Tesla Cash

On the global AI front, Chinese state media say DeepSeek and MiniMax have been approved to deploy their models internationally. It's a clear signal that Beijing wants homegrown AI competing abroad, which could mean tougher competition for American labs in markets outside the US.

China Clears DeepSeek, MiniMax Abroad

Meanwhile, OpenAI has fired three safety researchers. According to the report, the company's security team asked to inspect their personal phones and found disappearing Signal messages with an outside AI safety researcher. A week later, all three were let go, with one told it was a breach of trust. The episode shows the friction between tight secrecy at AI labs and researchers who want more openness about how these systems work.

OpenAI Fires Three Safety Researchers

Anthropic, for its part, has quietly done something unusual: it moved its main code repository off Git entirely. It built its own source control system called Scion and a review tool called Antique. The reason is striking. With AI-assisted development, daily code merges reportedly jumped from about 2,500 in May to more than 30,000 by September. The bottleneck stopped being writing code and became managing it.

Anthropic Replaces Git With Scion

That fits a comment from Google CEO Sundar Pichai, who said researchers are already using AI agents to help build new models. He was careful to add that truly self-improving AI is still a long way off.

Pichai on AI Building AI

Now to SpaceX, which is reportedly negotiating to lease at least 80,000 cell tower sites across the US, with Crown Castle among the companies involved. Combined with its Starlink satellites and newly acquired spectrum, the goal looks like a fourth nationwide wireless carrier. Traditional carriers' shares fell on the news, while tower companies may have found a big new customer.

SpaceX Pursues Fourth Wireless Carrier

And now, those dogs. A new company called Warhound has come out of stealth with 200 million dollars in funding. It was founded by Ben Lamm of Colossal Biosciences and Palmer Luckey of Anduril, and it wants to use gene editing and cloning to produce healthier, more capable working dogs for the military, police and rescue teams. Today many of these dogs are imported, and a large share wash out of training. Anduril will supply sensor-equipped gear. The first cloned dogs are expected early next year. It's ambitious, and certain to be controversial.

Warhound Gene-Edited Military Dogs

Staying with defense, L3Harris opened a new facility in Arkansas to consolidate solid-rocket motor production. It's expected to help double output for Lockheed Martin's GMLRS rocket artillery system, part of a wider push to expand US munitions capacity.

L3Harris Expands Rocket Motor Production

Some good news from medicine. Moderna and Merck report that their personalized mRNA cancer vaccine, given alongside Keytruda after surgery, cut the risk of melanoma returning or death by 42 percent in a late-stage trial. After two years, 82 percent of patients on the combination were alive without recurrence, versus 70 percent on Keytruda alone. The companies are already talking with regulators.

Moderna-Merck Melanoma Vaccine Results

For web folks, JPEG XL now works in both Chrome and Firefox. That clears the biggest hurdle for a format that handles high-quality and lossless images well, and can even shrink existing JPEG files without losing any quality.

JPEG XL Lands in Chrome, Firefox

Finally, two essays worth your time. Annie Vella argues that software engineers are becoming supervisors, directing AI, checking its work, and stepping in when needed, and warns they could lose skills if they drift too far out of the loop. And another piece makes the case that slow software is usually a design problem, not a language problem, so rewriting everything in a faster language rarely fixes much.

That's it for today's Automated Daily, tech news edition. The common thread is that AI is speeding everything up, including bug hunting, code output, and the bills, and the systems around it are struggling to keep pace. I'm TrendTeller. Thanks for listening, and I'll see you tomorrow.

PHP Exploits Hit Public Services & AI Uncovers Surge of Vulnerabilities - Tech News (Oct 23, 2026)

Our Sponsors

Today's Tech News Topics