Coding Tools
OpenAI Restricts Codex on Creature References
WIRED AIhigh confidence
Why it matters
- →This reflects OpenAI's efforts to enhance the utility and relevance of AI coding tools for developers.
OpenAI has instructed its coding agent, Codex, to avoid mentioning creatures like goblins and trolls unless directly relevant. This directive aims to streamline interactions and improve focus on coding tasks.
Read originalMore from WIRED AI
© WIRED AIMarket & Regulationbusiness
German Court Holds Google Liable for AI Errors
In a landmark decision, a German court has ruled that Google is liable for false statements generated by its AI Overviews feature. This ruling challenges the traditional view of search engines as mere conduits of third-party content, arguing that AI-generated summaries create new, independent statements. The court emphasized that Google's warnings about potential errors in AI outputs do not absolve it of responsibility, as these AI-generated claims can mislead users without any basis in the original sources. This decision could set a precedent for how AI-generated content is regulated globally, potentially impacting other tech companies using similar technologies.
WIRED AI
© WIRED AIMarket & Regulationmodels
Anthropic Takes AI Models Offline Due to US Order
Anthropic has removed its AI models, Claude Fable 5 and Mythos 5, from availability following a directive from the US government, which cited national security concerns. This action reflects ongoing friction between Anthropic and the Trump administration, which had previously labeled the company a 'supply chain risk.' The government order suggests a potential method to bypass the models' safeguards, though Anthropic maintains that the vulnerabilities are minor and not unique to their models. The situation highlights the complex relationship between AI development and regulatory oversight, raising questions about the transparency and fairness of such government interventions.
WIRED AI
© WIRED AIMarket & Regulationbusiness
Meta's AI Unit Faces Internal Turmoil
Meta's Applied AI team is embroiled in internal conflict, with employees voicing dissatisfaction over their roles and tasks. Formed to bolster AI research at Meta Superintelligence Labs, the unit is criticized for assigning tasks perceived as unchallenging and unfulfilling. This unrest is part of a larger morale issue at Meta following recent layoffs and restructuring efforts. CEO Mark Zuckerberg has acknowledged these difficulties and promised to provide more stability, but the situation underscores the ongoing tension between Meta's ambitious AI goals and the well-being of its workforce.
WIRED AI
More in Coding Tools
© GitHub ChangelogCoding Toolscoding
GitHub Copilot Code Review Enhances Customization
GitHub's Copilot code review has introduced new features that enhance customization and control for developers. Teams can now configure runner types at the organization level, allowing a single setting to apply across all repositories, which simplifies the setup process. The removal of the character limit on custom instructions provides more room for detailed guidance, while content exclusion settings ensure that Copilot respects organizational boundaries. These updates make Copilot code review more adaptable to specific team needs, offering greater flexibility and precision in code review processes.
GitHub Changelog
© The AI Daily BriefCoding Toolsproductivity
OpenAI Releases 'Sites' Feature in Codex
OpenAI has launched a new 'Sites' feature in Codex, enabling the creation of interactive, shareable documents.
The AI Daily Brief
Coding Toolsagents
Claude Code v2.1.172 Update Released
The latest update to Claude Code, v2.1.172, introduces significant improvements in agent functionality and user experience. Sub-agents can now spawn their own sub-agents up to five levels deep, enhancing automation capabilities. The update also addresses several bugs, such as session handling issues and model picker errors, ensuring smoother operation. Additionally, performance enhancements reduce idle CPU usage and improve tool loading in Chrome. These changes make Claude Code more robust and efficient for developers working with complex agent systems.
Claude Code Releases