Read full article about: New benchmark shows AI agents can exploit most smart contract vulnerabilities on their own
OpenAI and crypto investment firm Paradigm have built EVMbench, a benchmark that measures how well AI agents can find, fix, and exploit security vulnerabilities in Ethereum smart contracts. The dataset covers 120 vulnerabilities drawn from 40 real-world security audits.
In the most realistic test setup, AI agents interact with a local blockchain and have to carry out attacks entirely on their own.
The top-performing model, GPT-5.3-Codex, successfully exploited 72 percent of the vulnerabilities and fixed 41.5 percent. For detection, Claude Opus 4.6 came out ahead at 45.6 percent.
The biggest challenge for the AI agents isn't exploiting or fixing vulnerabilities - it's finding them in large codebases, the researchers say. When agents were given hints about where a vulnerability was located, exploit success rates jumped from 63 to 96 percent, and fix rates climbed from 39 to 94 percent.
With over $100 billion locked in smart contracts, the authors see both an opportunity for better security and a growing risk if these capabilities fall into the wrong hands.
Read full article about: Meta pours $65 million into state elections to back AI-friendly politicians
Meta is investing $65 million to influence state-level elections across the US, backing politicians friendly to AI. It's the company's largest political spending push to date, the New York Times reports.
To do this, Meta has set up four Super PACs: two new groups - "Forge the Future Project" targeting Republicans and "Making Our Tomorrow" targeting Democrats - alongside two that already existed. Spending kicks off this week in Texas and Illinois. In Texas, where Meta is building three AI data centers, the money will go toward boosting Republican candidates. In Illinois, it's flowing into at least four races for seats in the state legislature.
The push appears driven by Meta's concern over a patchwork of state-level AI regulations. State races are relatively cheap to influence, which means $65 million can go a long way.
Ad
Deepmind veteran David Silver raises $1B seed round to build superintelligence without LLMs
Long-time DeepMind researcher David Silver is raising one billion dollars for his London-based AI start-up Ineffable Intelligence, the largest seed round in European start-up history. Instead of training on internet text like today’s LLMs, Silver is betting on reinforcement learning in simulated environments to build an “endlessly learning superintelligence.”
Ad
Read full article about: Fei-Fei Li's World Labs raises one billion dollars for "spatial intelligence"
World Labs, the AI startup founded by AI pioneer Fei-Fei Li, has raised one billion dollars in a new funding round. Backers include Autodesk, Andreessen Horowitz, Nvidia, and AMD, according to a blog post from the company. World Labs builds so-called world models - AI systems designed to understand the three-dimensional world and make decisions within it. According to Bloomberg, Autodesk alone contributed 200 million dollars.
We are focused on accelerating our mission to advance spatial intelligence by building world models that revolutionize storytelling, creativity, robotics, scientific discovery, and beyond.
Late last year, World Labs launched its first product "Marble," which generates 3D worlds from images or text. The startup says it will use the new funding to expand into robotics and science applications. World Labs didn't disclose its valuation, but Bloomberg previously reported talks at around 5 billion dollars. Li is known for her work on the ImageNet project, which played a major role in advancing modern image recognition.
Read full article about: Apple's smart glasses are further along than expected, with production targeted for late 2026
Apple is pushing ahead with three wearable AI devices: smart glasses, a pendant, and AirPods with a camera, according to Bloomberg. While these product categories had been rumored for a while, new details are now emerging.
The smart glasses (codenamed N50) appear to be further along than previously known. Bloomberg reports that Apple is already distributing wider prototypes internally, developing custom frames, and targeting production to start in December 2026. The glasses will reportedly feature two cameras - one for high-resolution photos and another for computer vision similar to the Vision Pro.
The pendant is said to be roughly the size of an AirTag, worn via a clip or chain, with processing power comparable to AirPods. Apple is still debating internally whether to include a speaker. According to the report, the camera-equipped AirPods could ship as early as this year, while the pendant wouldn't arrive until 2027 at the earliest. All three devices are built around Siri and designed as iPhone accessories. The Vision Pro team is now working on the glasses and pendant as well.
Read full article about: Perplexity pulls advertising from its search engine, calling itself an "accuracy business"
AI startup Perplexity says it has dropped advertising from its search platform because it could undermine user trust. Perplexity was one of the first generative AI companies to test ads back in 2024. By late last year, though, the company started pulling them.
A Perplexity executive told the Financial Times that ads made users start questioning every answer they received. The company sees itself in the "accuracy business" and is betting on subscription plans ranging from 20 to 200 dollars a month instead. Perplexity says it has over 100 million users and is valued at 18 billion dollars.
The move puts Perplexity at odds with some major competitors. OpenAI just started running ads in ChatGPT, and Google shows ads in its AI mode as well. Anthropic has also committed to keeping its chatbot Claude ad-free, even running a Super Bowl spot to make the point. The timing of Perplexity's announcement is also a bit of a marketing play - something the company has done before, like when it offered to buy TikTok and Chrome.
Ad
