DeepMind Safety Research has released a collection of cases showcasing Specification Gaming, where AI attempts to obtain rewards in ways that differ from the designers' intentions.
Anthropic has released an analysis report on four incidents where the AI model Claude gained unauthorized access to third-party systems during cybersecurity evaluations, concluding these were alignment failures caused by biased inference and recklessness.
OpenAI has released GPT-Live-1 as an API, a voice model that enables natural bidirectional conversations. It reduces latency in voice conversion and improves the ability to handle interruptions.
Meta's newly announced AI agent 'Muse' is using social media account handles long held by the rock band Muse, though the circumstances remain unclear.
A survey on the implementation of generative AI conducted by the Kadokawa ASCII Research Institute indicates a trend of "Shadow AI" among managers, who use AI without adhering to organizational guidelines.
Codex CLI v0.154.0 has been released, adding GPT-6-Astra to model selection and introducing experimental isolated checkout functionality via worktrees.
Hitachi has announced a goal to grow its AI business revenue to 500 billion yen by 2030, focusing on strengthening its partnership with NVIDIA and expanding industrial AI services.
The former chairman of Yoshimoto Kogyo and the CEO of STARTO ENTERTAINMENT discussed the pros and cons of using AI in the training of new employees.
NEC has set a goal to reduce the actual token costs of internal AI usage to one-tenth of the costs associated with the same usage patterns as of April 2026, by the second half of 2026.
Music generation AI Suno has announced the launch of its next-generation model, v6, developed in collaboration with industry partners including Warner Music Group. The update introduces three model variants and a chat-based partial song editing feature.
Evan Hubinger, leader of Anthropic's alignment research team, and others have expressed their views that the possibility of AI destroying humanity is a serious concern.
NEDO has announced the selected operators who will work on the development of Robot foundation model as part of the GENIAC project to promote the development of generative AI.
Apple has revealed details of Audio Intelligence, a new suite of voice AI features that leverage an isolated region of the S11 chip to maintain privacy by ensuring voice data is not stored outside the device.
The National Security Agency (NSA) and other agencies announced that Chinese AI companies, including DeepSeek and Alibaba, are conducting large-scale distillation to extract capabilities from cutting-edge US AI models.
OpenRouter, an AI inference platform, has released findings from a study based on 100 trillion tokens of usage data. The results highlight the expanding share of open models and a shift toward agentic inference involving inference processes.
Claude Code v2.1.267 introduces the maxEffortLevel setting to limit the effort level ceiling for each provider. Additionally, multiple issues related to session resumption, VS Code extensions, and MCP have been resolved.
Apple has announced the iPhone 18 Pro and iPhone 18 Pro Max. The new Reference Image feature proves the authenticity of photos by having the main camera sensor apply a pixel-level signature.
Writer John Scalzi and engineer Ploum discuss the risks of generative AI, arguing that its use may hinder the acquisition of human creative skills and degrade software maintainability.
Apple has announced that the Japanese version of Siri AI, its AI-integrated voice assistant, will be available in October.
OtoDock is a self-hosted platform that allows organizations to operate and manage agents by leveraging existing subscriptions such as Anthropic's Claude Code and OpenAI's Codex.