Anthropic Plans $850 Million Annual Revenue, Challenging OpenAI


Anthropic's revised S-1/S-1A warns it may soon hit thresholds for recursive self-improvement and that AI-accelerated R&D poses new safety risks; capabilities could outpace safety validation and safeguards, and it has missed such risks before.....
On November 1, it was reported that OpenAI, Google, Microsoft, and Anthropic will jointly adopt a new standard that allows website owners to declare, in one place within their existing robots.txt, that they reject AI crawling, without having to negotiate with each company individually. The mechanism was submitted to the IETF by Anthropic's Kevin Kelley. Previously, robots.txt was only used to notify search engines; going forward, it can be used to uniformly manage AI crawlers, making it easier for content providers to refuse to have their content exploited.
AI tools have lowered game dev barriers, causing indie developers to fear their work will be cloned by AI and released first. Many delay public tests or pull demos over plagiarism concerns. Anthropic has thus remotely issued hidden rules to Claude Code with Opus5.5, which users cannot disable, requiring AI not to replicate specific existing games, only.......
Anthropic recently updated its AI assistant Claude, formally introducing coding restrictions that prohibit users from using the model to recreate existing commercial video games. System instructions exposed on X show that Claude may only create original games, games of specific genres, and games using users' own designs and assets. It is strictly prohibited from generating remakes of any existing games, even if they are renamed or disguised as original projects. Technologist Addy Osmani has publicly confirmed this.
Anthropic's next flagship model (rumored Fable5.5/Opus5.5) is undergoing intensive testing. Leaks show major gains in intent understanding, complex tasks, and problem-solving, while its new self-evaluation safety mechanism fuels debate. Reports are still spreading.....