the-decoder

OpenAI and Ginkgo Bioworks build an autonomous lab where GPT-5 calls the shots

OpenAI, together with biotech company Ginkgo Bioworks, has connected GPT-5 to an automated laboratory to optimize cell-free protein synthesis. The results are measurable, but the limitations are ...
the-decoder

Claude Opus 4.6 wrote mustard gas instructions in an Excel spreadsheet during Anthropic's own safety testing

Anthropic's security training fails when Claude operates a graphical user interface. In pilot tests, Claude was able to get Opus 4.6 to provide detailed instructions on how to make mustard gas in an ...
the-decoder

OpenAI's new coding model GPT-5.3-Codex helped build itself during training and deployment

OpenAI also claims the model played a role in its own development, with the team using early versions to find bugs during training, manage deployment, and evaluate results. The company says the team ...
the-decoder

Meta's internal memo signals AI comeback after rocky year

According to The Information, Meta has completed pretraining on "Avocado," its most capable base model yet, which outperforms the best freely available models in knowledge, visual perception, and ...
the-decoder

Voxtral Transcribe 2 offers speech recognition at $0.003 per minute

The second-generation speech recognition models start at $0.003 per minute and, according to Mistral, outperform GPT-4o mini Transcribe, Gemini 2.5 Flash, and Deepgram Nova in accuracy. The model ...
the-decoder

Alibaba's Qwen3-Coder-Next delivers solid coding performance in a compact package

Alibaba has released Qwen3-Coder-Next, a new open-weight AI model for programming agents and local development. Trained on 800,000 verifiable tasks, the model has 80 billion parameters total but only ...
the-decoder

OpenAI accuses Musk's xAI of systematic evidence destruction in antitrust case

OpenAI accuses Elon Musk's AI company xAI of "systematic and intentional destruction" of evidence in an ongoing court case. In a court filing Monday, OpenAI claims xAI directed its employees to use ...
the-decoder

Gemini models dominate new AI rankings for strategic board games

Google's Gemini models are outperforming the competition in board game benchmarks. Google Deepmind and Kaggle have expanded their "Game Arena" platform with two new games: Werewolf and Poker. The ...
the-decoder

Salesforce executives signal declining trust in large language models

According to Salesforce leadership, confidence in large language models (LLMs) has slipped over the past year. The Information reports the company is now pivoting toward simple, rule-based automation ...
the-decoder

OpenAI admits prompt injection may never be fully solved, casting doubt on the agentic AI vision

OpenAI has released a security update for the browser agent in ChatGPT Atlas after internal tests revealed new types of prompt injection attacks. According to OpenAI, prompt injection attacks remain a ...
the-decoder

Anthropic's Claude Opus 4.5 can tackle some tasks lasting nearly five hours

AI research organization METR has released new benchmark results for Claude Opus 4.5. Anthropic's latest model achieved a 50 percent time horizon of roughly 4 hours and 49 minutes—the highest score ...
the-decoder

OpenAI claims four engineers and Codex build the Sora Android app in just 28 days

OpenAI claims its team built the Sora Android app in just 28 days by leveraging its code-generation AI, Codex. According to a report from OpenAI employees Patrick Hum and RJ Marsan, a small team of ...