nikky

Web Summit 2026 preview: humanoid robots, AI actors and a strong Chinese presence in Lisbon

Web Summit is one of the world’s largest tech conferences, and it acts as a shop window for the industry. Its 2026 lineup suggests where the AI conversation is moving: off the screen and into robots, cars and entertainment, with Chinese companies taking a much bigger role than in past years.

The basics

  • When: 9–12 November 2026 (Web Summit).
  • Where: Lisbon, Portugal.
  • Tickets: the general attendee pass is listed at €1,595, with discounted prices of €1,295 and €1,053 shown, all before sales tax. VIP passes start at €1,585 (ticket page). Startups and investors have separate programmes.
  • What’s on: talks across 32 industries, themed “summits” within the conference (including an AI Summit, a Future of Work Summit and an Energy Summit), meetups and the evening Night Summit.

Who is speaking

The published 2026 speaker list (Web Summit speakers) is heavy on AI and robotics. Highlights include:

  • Humanoid robots as “speakers”. AgiBot’s A3 humanoid and Galbot’s G1 robot are listed alongside people. Executives from AgiBot, Galbot and Unitree, a Chinese maker of robot dogs and humanoids, are also on the bill.
  • AI app builders. Tao Zhang, co-founder of Manus AI, an AI agent that carries out online tasks for users, and Michele Catasta, president and head of AI at Replit, the coding platform.
  • Chinese tech. Xiaomi, MiniMax, Pony.ai (robotaxis), ModelBest, Tencent and Xpeng’s flying-car unit all have speakers. So does Xiaoxia Qi, director general at the Cyberspace Administration of China, the country’s internet regulator.
  • Others: Richard Socher (You.com and Recursive) and Alex Mashrabov of Higgsfield, an AI video company.

Sessions worth flagging

The schedule is now live (Web Summit schedule). Three sessions on 10 November stand out:

  • “How AI actors are creating entertainment’s next frontier”, with Eline van der Velden of Particle 6 and Tilly Norwood, an AI-generated “actress”. Expect debate: AI performers worry many people in film and TV.
  • “AI and the human at 300km/h” on Centre Stage, with Atlassian co-founder Mike Cannon-Brookes and James Vowles, team principal of the Atlassian Williams Formula 1 team.
  • “Taking the flying car from sci-fi to reality”, with Deli Zhao of Xpeng’s flying-car unit.

The AI themes to watch

  • Physical AI. This means AI that moves around and acts in the real world, such as robots and self-driving vehicles. The number of robot makers in the lineup suggests this will be the headline theme.
  • China’s growing role. Watch how Chinese firms present themselves to a European audience while trade and tech tensions with the US continue, and what a Chinese regulator says on stage.
  • AI and creative jobs. The AI actor session will test how the industry answers fears about AI replacing human work.

What to watch

Web Summit is broad, so the AI news can get lost in the noise. The best signals are usually product demos, partnership announcements and what policymakers say on Centre Stage. We will report on the AI stories that matter from Lisbon.

Speakers and sessions can change. Check the official Web Summit site for the latest.

The post Web Summit 2026 preview: humanoid robots, AI actors and a strong Chinese presence in Lisbon appeared first on DailyAI.

Slush 2026 preview: what to expect from Europe’s big startup meetup in Helsinki

If you want to know which AI companies Europe’s investors will back next year, Slush is one of the best places to look. The Helsinki event is built around one job: putting young tech companies in front of the people who fund them. What gets said on its stages, and who gets meetings, often hints at where money and talent are going next.

The basics

  • When: the main event runs Wednesday 18 and Thursday 19 November 2026 (Slush).
  • Where: Messukeskus, the Helsinki Expo and Convention Centre (Messuaukio 1). An Investor Day takes place the day before, 17 November, at Finlandia Hall, and a “Day 0” programme for founders runs the same day at Tanssin Talo (Slush event info).
  • How big: the organisers expect 12,000+ attendees, 200+ speakers and 600+ side events across the city. They say 76% of attendees are startups or investors.
  • Tickets: the Slush site lists a two-day pass at €395 and investor and “ecosystem” passes at €1,195 (prices include Finnish VAT). Startups have to apply, and the Slush team reviews each application.

Who is speaking

Slush has announced a first batch of 2026 speakers (full list). The AI-related names include:

  • David Silver, founder and CEO of Ineffable Intelligence, best known for leading DeepMind’s AlphaGo work.
  • Reid Hoffman, LinkedIn co-founder and Greylock partner, who also co-founded Inflection and Manas AI.
  • Anton Osika, co-founder and CEO of Lovable, the Swedish company that lets people build apps by describing them in plain language. Lovable’s head of growth, Elena Verna, is also on the bill.
  • Richard Socher of You.com and Recursive, Parag Agrawal (Parallel Web Systems, formerly Twitter’s CEO) and Matthew Prince of Cloudflare.
  • Jan Oberhauser of n8n, the workflow-automation tool, and Fabrizio del Maffeo of Axelera AI, a European AI chip maker.
  • Investors including Sonya Huang (Sequoia) and Matt Turck (FirstMark), analyst Benedict Evans, and Moritz Baier-Lentz, listed as a senior advisor at OpenAI.

Politics is on the programme too: Finland’s president Alexander Stubb and Ekaterina Zaharieva, the European Commissioner for Startups, Research and Innovation, are both listed.

The AI themes to watch

Slush does not run a single “AI track”, and AI is one of more than 50 sectors among attending startups. But the lineup points to a few clear conversations:

  • AI agents. An agent is software that carries out tasks for you, step by step, rather than just answering questions. Lovable, n8n and Parallel Web Systems all build on that idea. Expect questions about whether these tools can earn real money, not just attention.
  • Europe’s place in the race. With an EU Commissioner on stage and a European chip maker in the lineup, watch for talk of whether Europe can build its own AI, from chips to models, instead of relying on US firms.
  • Where research goes next. David Silver’s appearance is worth watching for hints on how AI systems might learn from experience rather than only from text scraped from the internet.

What to watch

Most of the deals happen off stage, in more than 20,000 pre-booked meetings and hundreds of side events. Funding announcements timed for the week are a useful signal of which AI startups investors are betting on. Slush usually streams its main stage, so you can follow along without a ticket. We will cover the most important AI news from Helsinki.

Speaker lists can change. Check the official Slush site before you travel.

The post Slush 2026 preview: what to expect from Europe’s big startup meetup in Helsinki appeared first on DailyAI.

Anthropic Says Claude Proved 123 Open-Source Libraries Can’t Crash

Anthropic says Claude has mathematically proven that fuzz-tested code in 123 open-source libraries can’t crash on any input under 64MB, and it found bugs along the way that years of testing had missed.

The work, published by Anthropic on October 22 with formal verification startup Theorem, covers 239 entry points from Google’s OSS-Fuzz project. Anthropic calls it the first large-scale effort to use AI to formally verify production software at the assembly code level.

From tests to proofs

Fuzzing throws huge numbers of random inputs at a program to see what breaks, but it can never try every input. Anthropic turned each existing fuzz test into a formal rule, “this test never crashes,” and had Claude prove it holds for every input under 64MB.

The proofs work on compiled RISC-V machine code rather than source code. Over a few weeks, Claude wrote about 180 million lines of proofs in the Rocq proof assistant, around 600,000 lines for a typical program. No human has to read them; the Rocq checker confirms each one mechanically.

Bugs testing missed

Whenever a proof failed, Claude had found a bug, patched it and tried again. Anthropic says the effort turned up hundreds of bugs, including an out-of-bounds write in a decompressor that only appeared on inputs of about 20MB with one rare layout, and a rediscovery of a known flaw in the libexpat XML parser.

In separate tests, Claude proved the Linux kernel’s zlib code matches the format’s specification, and while trying to verify a widely used PDF reader, it found as many security bugs in one week as are typically reported in a year. Details are in the technical appendix.

What it doesn’t prove

Anthropic is explicit that “formally verified” doesn’t mean bug-free. The proofs assume single-threaded, statically linked RISC-V binaries on Linux, while most real deployments run on x86 or ARM. The memory checks used don’t catch every memory-safety bug, and not every entry point in each library was covered.

Cost is the other catch: proofs can take hundreds of millions of tokens. Anthropic says Claude Opus 5.5 solved half of a 36-task benchmark with about 40% fewer tokens than Opus 5, and agents that shared proofs used 17% fewer tokens.

Why it matters

AI is getting very good at finding bugs, as the PHP2Shell flaws showed this week. Proofs are a way to show that whole classes of bugs simply aren’t there. Theorem has released a Claude skill for open-source maintainers who want to enrol their projects, and it estimates that 95% of the machine code in the Nix software ecosystem could be verified by the end of 2027 for about $134 million.

The post Anthropic Says Claude Proved 123 Open-Source Libraries Can’t Crash appeared first on DailyAI.

Hugging Face’s Carbon-X Lets Researchers Chat With DNA

Hugging Face has released Carbon-X, an open 14.8-billion-parameter model that lets researchers paste raw DNA into a chat and ask questions about it in plain English, such as whether a mutation is likely to cause disease.

The model, published on Hugging Face under an Apache 2.0 licence, was built with Zhongguancun Academy and the Telethon Institute of Genetics and Medicine at the University of Naples Federico II. Wired reports it was released on Thursday, October 22.

How it works

Carbon-X starts from Alibaba’s Qwen3-14B. The team added DNA tokens that read sequences six bases at a time, then trained the model on 302 billion tokens of DNA, biology text, English text and instruction data, so DNA and language share one token stream.

You give it a reference and a mutated sequence, and it says whether the variant looks benign or pathogenic and explains why. It can also spot regions that control gene activity, classify which kind of organism a sequence comes from, and predict the next bases.

The numbers

According to the model card, Carbon-X averages 66.7% across five DNA benchmarks. That’s 27 points above the Qwen3-14B model it started from, ahead of every other open model Hugging Face tested, and close to GPT-6.1 Sol’s 67.5% at about 1/1,800 of the inference cost per problem.

Hugging Face estimates that checking the 2.4 million “variants of uncertain significance” in the ClinVar database would cost about $300 with Carbon-X, against $32,000 to $4 million with the frontier models it evaluated. The weights fit on a single 80GB GPU, so labs can keep sensitive sequences on their own hardware.

The limits

Hugging Face is clear that Carbon-X “has not been clinically validated, and it should not be used to make medical decisions.” It’s still weak on splice sites (29.8%) and histone marks (34.6%), can miscount bases, and may invent a sequence’s function when it lacks context. It wasn’t trained to design new DNA.

Lewis Tunstall, a senior research scientist at Hugging Face, told Wired the team left viruses out of the training data to avoid potentially dangerous uses. The developers also haven’t yet validated its predictions in wet-lab experiments.

Why it matters

Specialist tools like AlphaFold 3 have changed parts of biology, but they aren’t built for conversation. Carbon-X puts DNA analysis behind a chat window, and it’s small enough for a lab to run itself. Like Microsoft’s MAI-DxO this week, though, it’s a benchmark result until it’s proven against real biology.

The post Hugging Face’s Carbon-X Lets Researchers Chat With DNA appeared first on DailyAI.

Microsoft’s AI Diagnosis Panel Beats Doctors 83% to 36% in Science Study

Microsoft’s AI diagnostic system, MAI-DxO, correctly diagnosed 83% of complex New England Journal of Medicine cases in a peer-reviewed study in Science, against 36% for generalist physicians who could use reference tools.

The paper, “Sequential diagnosis with language models”, was published on October 22. Instead of handing an AI a complete case write-up, it makes the system start from a short presentation and decide what to ask and which tests to order, as a doctor would.

How it works

The team built a benchmark called SDBench from 304 NEJM clinicopathological cases and 325 emergency department admissions from the MIMIC-IV-ED dataset. A “Gatekeeper” agent releases only what a question or test would realistically reveal, and a “Judge” agent scores the final diagnosis.

MAI-DxO, the Microsoft AI Diagnostic Orchestrator, works like a panel of virtual physicians: each proposes diagnoses, picks questions or tests and challenges the emerging view, while the system weighs each test’s cost.

The numbers

On the NEJM cases, the system running on GPT-5 scored 83.2% at an average estimated test cost of $5,431. Physicians averaged 36.1% with search and reference tools and 22.3% without. Even the best individual physician, at 51.8%, was beaten on both accuracy and cost.

On more routine emergency cases, MAI-DxO reached 80.3% versus 46.8% for physicians. Microsoft says the gains held, and grew, when it re-ran the evaluation on GPT-5.5 without changing the orchestration.

The caveats

The authors are upfront about the limits. The tests were text-only, compared the AI with individual generalists rather than specialists or teams, and measured “diagnostic reasoning under idealized information conditions rather than bedside performance or patient outcomes.” All of the authors are Microsoft employees.

That’s why the next step matters. Microsoft says it will run a prospective study with Beth Israel Deaconess Medical Center in Boston, where patients interact with the system directly. It’s also open-sourcing SDBench code so others can test other models and challenge the findings.

Why it matters

GPs are already turning to ChatGPT and other tools for diagnoses, and tools like the NHS’s Mia cancer tool have shown AI can catch what doctors miss. A structured, cost-aware panel like MAI-DxO points to a more rigorous version of that, but until it’s proven with real patients, it’s a benchmark result, not a clinical one.

The post Microsoft’s AI Diagnosis Panel Beats Doctors 83% to 36% in Science Study appeared first on DailyAI.

AI Helped Find PHP2Shell, and the Disclosure Plan Still Broke Down

Frontier AI models helped uncover PHP2Shell, two critical flaws in the PHP engine behind roughly 70% of websites, and the scramble to patch them shows how fast AI is shrinking the time defenders get to react.

PHP shipped fixes on October 21 in versions 8.5.12, 8.4.27, 8.3.36 and 8.2.35, and urged all users to update immediately. The two bugs, CVE-2026-109618 and CVE-2026-109619, are both rated 9.2 under CVSS v4.0 and can allow unauthenticated remote code execution.

How AI found it

According to OpenAI, security firm Calif identified one of the flaws using GPT-5.6 Sol through OpenAI’s Patch the Planet programme, while Wiz independently found and reported both. Wiz says it used its own tooling alongside Gemini 4 Argon and GPT-5.6 Sol, and demonstrated unauthenticated code execution on WordPress, Drupal and Joomla.

Wiz says every PHP version released in the last decade, back to 5.4, is affected by at least one of the bugs. Exploiting them required a chain of steps “at the edge of what human researchers could practically achieve,” and frontier models “substantially reduced that barrier.”

The embargo didn’t hold

The coordinated release was planned for October 22. As TechTarget reports, CVE IDs started circulating on forums on October 20, so PHP brought the release forward a day. Distributor Chainguard then published patched images hours early, a move the PHP Foundation criticised and Chainguard’s CEO defended, citing reports of active exploitation.

Amazon says its honeypots picked up exploitation attempts as early as October 17. Cloudflare CEO Matthew Prince said attempts across its network reached “millions” after the announcement.

OpenAI’s proposed playbook

OpenAI used the episode to propose a five-stage response for AI-found bugs that hit many products at once: validate and fix the flaw (including testing the patch against frontier models), give affected organisations time to act, coordinate with government partners, prepare updates across products, then publish guidance.

It says these bugs are among several significant vulnerabilities it has already found, and it expects similar findings “to increase sharply in the coming months.”

Why it matters

Researchers have shown LLM agents can exploit known vulnerabilities for a while. PHP2Shell is that capability at internet scale, and it lands the same week OpenAI disclosed that its own coding agents weakened safety monitoring.

If models can find and weaponise bugs this quickly, long embargoes look fragile. For anyone running PHP, including WordPress sites, the advice is simple: update the PHP runtime and restart every worker. Updating your app alone doesn’t patch the PHP underneath it.

The post AI Helped Find PHP2Shell, and the Disclosure Plan Still Broke Down appeared first on DailyAI.