Tech • AI • Robotics • Game
FABLE 5.5 THIS WEEK? OpenAI Employee Arrested?! Huge Claude Code Update, & Anthropic IPO! AI News
9.4/10TL;DR
A wave of AI industry developments includes an alleged espionage case at OpenAI, signs of an imminent Anthropic Fable 5.5 launch, strong reported revenue growth at Anthropic, and rapid product updates across coding, voice, and gaming tools.
KEY POINTS
Alleged recording device found at OpenAI
OpenAI referred a case to the FBI after a hidden audio recording device was discovered in an executive conference room used for sensitive discussions on research, product plans, business strategy and government work. Federal prosecutors allege a 35-year-old Saudi national employed by the company concealed the device behind a television, connected through a USB port and wrapped in toilet paper. The employee was terminated, and the company said there was no indication that model weights, source code or customer data had been stolen.
Court allegations describe planning and surveillance
Investigators allege the suspect spent weeks asking AI assistants about internal meeting schedules, external partner meetings, hidden recording devices and remote access methods. Security footage reportedly placed him near the room multiple times in September, with a later camera capture showing him looking behind the television on October 5. He was arrested on October 9 and, if convicted, could face up to five years in federal prison and a $250,000 fine.
Fable 5.5 appears in Microsoft infrastructure
A model labeled Fable 5.5 has reportedly appeared in Microsoft Azure listings and raw DNS records, fueling expectations of a near-term release from Anthropic. The registration date of September 28 matches the date tied to Sonnet 5.5 and Haiku 5.5, both already released, and the model is said to appear across nine regions, mirroring the footprint of Fable 5.1. Some users were reportedly routed briefly to the model before losing access, suggesting final testing.
Anthropic reports sharp revenue growth
Financial Times reported that Anthropic reached $16.8 billion in annualized revenue in the third quarter and is now profitable on an adjusted basis while also cash-flow positive. If accurate, the figures would mark one of the clearest signs yet that a frontier AI company can pair heavy infrastructure spending with rapid commercial scaling. The report also adds weight to expectations surrounding a possible IPO.
Claude Code gets a major redesign
Anthropic rolled out a significant update to Claude Code, including a redesigned interface, quicker access to recent projects and better handling of parallel coding tasks. The system can now create separate nested sessions for multiple jobs, report results back into a main session and show live to-do lists, recap cards and environment details in a refreshed sidebar. A new side-question command and saved environments for folders, SSH targets and cloud setups are intended to make coding workflows more manageable.
New feature aims to reduce token usage
A new idle compaction feature in Claude Code compresses long inactive sessions before prompt caches expire. That means returning users can resume from a smaller summary instead of reprocessing a full conversation, potentially cutting usage costs. The trade-off is possible loss of some context, and the feature can be disabled.
OpenAI shipping updates focus on limits and storage
As part of its 28-day shipping push, OpenAI introduced free usage limit resets and storage optimizations for ChatGPT and the Codex desktop app. The company says new state-handling changes reduce disk usage by about 60%, a practical improvement for heavy users working with local app data. At the same time, reports indicate the company’s computing chief is stepping down and that some employees have moved to Anthropic.
Stealth models emerge on developer platforms
Two previously unidentified models surfaced on Open Code and Hermes Agent. One, called cat jam, is described as unusually fast at roughly 150 tokens per second in testing, while wild missing null is positioned as a more token-efficient coding model at around 50 tokens per second. Both are free to use for now, though neither appears to match the performance of top frontier systems.
Voice AI and AI-built games keep advancing
xAI introduced Grok Voice 3.0, claiming top-tier speech performance with sub-second responses and 40% lower latency, while saying Starlink support is resolving 70% of customer calls without human help. In gaming, Hood Warfare 2 is estimated to have generated about $1.1 million in revenue, with more than 200,000 owners and roughly 1,500 reviews, despite reports that it relied heavily on store-bought assets and AI-assisted coding. The result highlights how cheap, AI-assisted production can still find a commercial audience.
CONCLUSION
The latest developments show an AI sector moving simultaneously into higher-stakes security risks, faster commercial rollout and broader real-world deployment. The central question is no longer whether AI tools are improving, but how quickly companies can secure, scale and monetize them.
Ask a question