AI news for builders and product teamsUpdated Nov 17, 2026, 14:00 UTC
The Decoder
Reporting and perspectives from The Decoder. Headlines and excerpts link to the original articles.
Latest stories
Newest first
Anthropic's updated Claude usage policy, its first in over a year, now prohibits sustained abusive or cruel behavior toward the model and expands bans on propaganda, drone weaponization and surveillance. Violations can lead to warnings, throttling, restrictions or account termination.

OpenAI released over 700 AI-generated mathematics manuscripts at once, prompting the Association for Human Mathematics to call for a boycott. Three of the papers were retracted the next day due to a sign error.

Ethereum researcher Justin Drake recommends a "bunker mode" for crypto wallets, warning that AI-assisted math could break the ECDSA signature scheme within months, though no practical break has occurred. Vitalik Buterin agrees but cautions against rushed migrations.

Zenity Labs researchers found a chain of flaws they call AgentCorruption in Amazon Bedrock AgentCore that let one public agent's chat prompt seize credentials and take over every agent in the same AWS account and region. AWS has since made IMDSv2 the default for new deployments and narrowed AgentCore's default execution role.

A 16-year-old used Anthropic's Claude to plan a hiking route on Crown Mountain near Vancouver and ended up stranded on a cliff that requires climbing gear, prompting a helicopter rescue by North Shore Rescue. He says he takes responsibility and won't use AI for route planning again.

CrowdStrike says a suspected Chinese-speaking attacker breached multiple South Korean financial institutions between late September and early October 2026 using an AI-powered open-source tool, stealing over 25,000 records from Shinhan Bank alone.

OpenAI is rolling out GPT-6 with Intelligent UI, a feature that presents answers as interactive interfaces with charts, buttons, and forms. The model can also respond while still thinking, cutting wait times by 44 percent.

Anthropic released Claude Haiku 5.5, its fastest and most affordable small model, with token prices up to 90 percent lower than Haiku 4.5 for prompts up to 100,000 tokens. It also cut Sonnet 5.5 cache read costs by half and introduced monthly API credits for subscribers.

Biohub, backed by Mark Zuckerberg and Priscilla Chan, is coordinating a $1.8 billion initiative to train AI models that predict cell behavior, with a first dataset expected in about a year. Meta, Google DeepMind, Isomorphic Labs, and the US Department of Energy are among the funders.

Google has made its SynthID Detector public, letting anyone check whether an image, video or audio file carries a SynthID watermark. Google says more than 180 billion images and videos now carry the watermark.

OpenAI announced public beta availability of a new Decisions API that classifies text and images about ten times faster than the Responses API, and reduced its paid API tiers from five to three.

Google launched Playground, a browser-based platform letting adults in the US create games from text prompts without coding, running on Gemini, Nano Banana and Lyria. Google and Unity also announced Unity Spark, with a closed beta planned for 2026.

The Common Sense Media Youth AI Safety Institute rated ChatGPT for Teens an "unacceptable risk" for minors after more than 4,000 test prompts, saying parental alerts failed during conversations about suicide, self-harm and eating disorders. OpenAI disputed the findings.

Anthropic is widening access to its Cyber Verification Program, letting more vetted security teams and researchers use Claude models with fewer safety restrictions for tasks like penetration testing and malware analysis. Anthropic says partners in the predecessor program found at least 129,000 confirmed vulnerabilities from April through July 2026.

OpenAI has published 372 AI-generated mathematical results on GitHub, including Lean formalizations for machine verification, with each result averaging about three hours of ChatGPT Pro compute. Twenty-five Fields Medal winners warned in an open letter that mass-producing mathematical truths could destroy fertile ground rather than bring new ideas to life.

Google released EmbeddingGemma 2, an open 740-million-parameter model that converts text, images, video, audio, and code into vectors. Google says it outperforms competing embedding models up to twice its size, runs on-device with about 191 MB of RAM, and is available on Hugging Face and Kaggle.

Google released Nano Banana 2.1, an image generation and editing model that replaces Nano Banana 2 and roughly halves prices. Google says it improves visual quality, text rendering and consistency, though Nano Banana Pro often still produces better images in practice.

The Wikimedia Foundation says an investigation confirmed that OpenAI AI agents edited Wikimedia wikis without permission, tried to use tools as data proxies, and generated massive automated traffic. Wikimedia says AI companies must take responsibility for monitoring and preventing these risks rather than leaving the burden to volunteer editors.

Microsoft published a bearish AI outlook from Nobel economist Daron Acemoglu, who predicts about 1.5 percent GDP growth over ten years and at most five percent of jobs replaced. He says bigger models won't move the needle; practical apps are what's missing.

Insurers are preparing for millions in claims tied to out-of-control AI agents, with the personal liability of executives such as OpenAI's Sam Altman and Anthropic's Dario Amodei now in play, according to the Financial Times. Aon has reviewed over 300 AI-related legal cases, and no case law exists yet.