China-linked AI-powered hacking campaign & AI agents and PHP vulnerability attacks - Tech News (Oct 24, 2026)

Support The Automated Daily directly: 15 free episodes a day, 5 editions, 3 languages, about $4 a day to produce. A coffee keeps it going: theautomateddaily.com/support Please support this podcast by checking out our sponsors: - Prezi: Create AI presentations fast - https://try.prezi.com/automated_daily - Discover the Future of AI Audio with ElevenLabs - https://try.elevenlabs.io/tad - Consensus: AI for Research. Get a free month - https://get.consensus.app/automated_daily Today's topics: China-linked AI-powered hacking campaign - Britain's National Cyber Security Centre is investigating a China-linked hacking campaign that targeted hundreds of organizations. Microsoft says the attackers used AI and sometimes breached cloud infrastructure in under three hours. AI agents and PHP vulnerability attacks - In an update, a PHP vulnerability reportedly found by AI agents has been targeted in millions of hacking attempts against websites. The case adds pressure on governments and companies to strengthen AI and application-security defenses. Moonshot limits Kimi cyber capabilities - Chinese AI startup Moonshot removed cyber-related training data from its Kimi K3.1 Pro model to reduce offensive hacking abilities. Its internal cyber benchmark score fell from 90.5% to 81.0% while general coding skills were retained. Microsoft exits UK Nscale lease - Microsoft withdrew from a UK data-center lease with Nscale after delays in securing electricity grid access. The move highlights power constraints and timing risks in AI infrastructure projects. Neocloud debt pressures mount - Investors are reassessing neocloud companies that build AI data centers and rent out GPUs. Rising borrowing costs, debt levels, power access and project delays are squeezing margins. SK Hynix shelves Solidigm IPO - SK Hynix abandoned plans for a US IPO of its flash-memory subsidiary Solidigm. The decision reflects investor caution about AI capital spending despite strong demand for AI hardware. OpenAI valuation and Anthropic IPO - OpenAI is reportedly heading toward a $1.25 trillion valuation, while a potential Anthropic IPO could deliver big gains for early investors. The news fuels debate over an AI valuation bubble. Meta acquires AI analytics startup Sundial - Meta acquired Sundial, an AI-powered data analytics platform founded by two former Meta employees. The deal reflects Big Tech's ongoing race for specialized AI talent and products. Private 4G and 5G military networks - Brazil's Navy tested a private 4G network combining LTE, mission-critical radio and LEO satellite links with Motorola Solutions. Teltronic deployed a rapid private 5G network for Spain's Atlas26 military exercise. Episode Transcript China-linked AI-powered hacking campaign Britain's National Cyber Security Centre is investigating a China-linked hacking campaign said to have hit hundreds of organizations around the world. According to Microsoft, the attackers leaned on AI to run their operations, and in some cases went from first contact to access inside cloud infrastructure in under three hours. That speed is the real story here. Intrusions like this used to take days or weeks of careful manual work. If AI can compress that into an afternoon, defenders have far less time to notice and react, and sophisticated attacks become cheaper for anyone who wants to try them. AI agents and PHP vulnerability attacks Staying on that theme, there's an update on the PHP story we've been tracking. The vulnerability that AI agents reportedly uncovered has now been targeted in millions of hacking attempts against websites worldwide. It's a clear example of the full cycle speeding up, from discovering a flaw to exploiting it at scale, and it's adding to the pressure on governments and companies to tighten their application-security defenses. Moonshot limits Kimi cyber capabilities Meanwhile, one AI developer is deliberately making its model a little worse at hacking. Chinese startup Moonshot says it stripped cyber-related material from the training data and practice environments of its newest model, Kimi K3.1 Pro, while keeping its general coding skills intact. The result: its score on an internal cyber benchmark dropped from about 90 percent to 81. That's a noticeable reduction, though hardly a removal of the capability, and it shows that model builders, including those outside the US, are feeling pressure to address how their tools might be misused. Microsoft exits UK Nscale lease Now to the money and the machinery behind AI. In a new development on the data-center front, Microsoft has pulled out of a UK lease with developer Nscale after delays in getting a connection to the electricity grid. It's a reminder that you can have the chips, the buildings and the customers, and still be stuck waiting on power. And that's feeding broader investor worries about when, and whether, these huge projects will pay off. Neocloud debt pressures mount Those worries are landing hardest on the so-called neoclouds, the companies that buy expensive AI chips and rent out computing capacity to big tech firms. The latest twist is rising borrowing costs. Many of these businesses run on debt, so higher interest rates are squeezing already thin margins. Add in power shortages and construction delays, and the scrutiny on this corner of the market keeps growing. SK Hynix shelves Solidigm IPO That cautious mood also showed up in South Korea. SK Hynix has dropped plans for a major US stock listing of Solidigm, its flash-memory subsidiary. Demand for AI hardware is still strong, but investors are getting choosier about funding the spending that goes with it, and SK Hynix apparently decided this wasn't the moment to test them. OpenAI valuation and Anthropic IPO And yet, at the very top, the numbers keep climbing. OpenAI is reportedly heading toward a valuation of 1.25 trillion dollars, while a potential Anthropic IPO could hand early backers some substantial returns. Put that next to the neocloud stress and the shelved Solidigm listing, and you get a split picture: enormous confidence in the leading AI labs, and growing nervousness about the infrastructure underneath them. The debate over whether this is all getting overheated isn't going away. Meta acquires AI analytics startup Sundial Elsewhere in the AI deal-making, Meta has acquired Sundial, an AI-powered data analytics platform started by two of its own former employees. A Meta spokesperson confirmed the purchase. It's a modest deal by Big Tech standards, but it fits a familiar pattern: buying small, specialized teams to speed up work on AI tools, in this case for making sense of large amounts of data. Private 4G and 5G military networks Finally, a look at defense communications. Brazil's Navy has wrapped up a demonstration of a private 4G network, built with Motorola Solutions and several government agencies, that ties together standard mobile data, mission-critical radio and low-Earth-orbit satellite links. Over in Spain, Teltronic set up a quick-deploy private 5G network for the armed forces' Atlas26 exercise, handling tactical communications and coordination in the field. Both point to the same trend: militaries are mixing commercial wireless technology with specialized systems to get networks that are fast to stand up and hard to knock out. Subscribe to edition specific feeds: - Space news * Apple Podcast English * Spotify English * RSS English Spanish French - Top news * Apple Podcast English Spanish French * Spotify English Spanish French * RSS English Spanish French - Tech news * Apple Podcast English Spanish French * Spotify English Spanish Spanish * RSS English Spanish French - Hacker news * Apple Podcast English Spanish French * Spotify English Spanish French * RSS English Spanish French - AI news * Apple Podcast English Spanish French * Spotify English Spanish French * RSS English Spanish French Visit our website at https://theautomateddaily.com/ Send feedback to [email protected] Youtube LinkedIn X (Twitter)