All articles tagged with #claude
In a comprehensive side-by-side test, ChatGPT outshines Claude in deep research, image generation, and overall usability, while Claude offers larger context windows and strong HTML-based visualizations; both have free and paid tiers with differing feature access, but the verdict favors ChatGPT as the more versatile all-around option, though users may prefer Claude for specific needs.
Anthropic is preparing to publicly file its IPO paperwork as soon as this month, aiming to go to market ahead of rivals like OpenAI. The filing would disclose financials, revenue by segment, and details of its Claude-based products and enterprise offerings following a high-value funding round that boosted its valuation toward the trillion-dollar mark.
Anthropic says Claude will apply invisible watermarks using a SynthID-Text method to comply with the EU AI Act’s requirement that synthetic text be machine-readable and detectable as generated content. The watermarking uses a key and surrounding words to influence word choices, remains invisible to readers, and does not affect Claude’s output quality or cost. Google’s Gemini already uses SynthID-Text, and while OpenAI’s ChatGPT plans aren’t detailed, it will also fall under the same regulations.
Anthropic detailed Claude’s text watermarking system, based on the SynthID-Text approach, which uses low-stakes word choices to imprint an undetectable pattern that signals AI involvement without affecting output, extra tokens, or reader experience; a forthcoming detection API will let users check text, but the watermark is not a foolproof test for human vs. AI authorship and works best on longer passages, with reduced effectiveness on factual or short samples.
Anthropic faced a major outage across Claude.ai, Claude Code, and Claude Cowork starting with authentication issues and degraded performance; all services were restored by 22:40 UTC the same day, while Claude Console and the API remained operational.
Ruben Circelli built FISH, an AI-powered Finals study helper that converts pro YouTube guides into random, in-game drills, simulates complete matches, and offers both rulebook and pro-strategy advice; it also quizzes memorization of gadgets, maps, and icons. FISH leverages Opus 5 and Claude to distill a strategy bible from transcripts, then applies it to generated states. Opus 5 is fast and capable but can burn through a five-hour limit and struggles with some web data extraction and memory-following precision. After a week of use, Circelli reports clearer understanding and faster decision-making, though GPT-5.6 still often performs best on other tasks.
Anthropic detailed its Claude watermarking system, rooted in a secret-key token approach inspired by Google’s SynthID, claiming negligible speed impact and that watermarking could appear even when Claude edits text; the watermark is intended to help comply with the EU AI Act. Despite assurances, some Claude users on X report canceling subscriptions in protest, though Anthropic says overall cancelations have not spiked.
Anthropic reported preliminary Q2 revenue of more than $11.5 billion—more than 14 times the year-ago quarter—driven by demand for its Claude AI software, with positive adjusted operating income and a run-rate of about $47 billion. As it vies with OpenAI for enterprise customers, Anthropic is eyeing an IPO as soon as this fall to secure capital for expanding compute infrastructure and data centers.
Anthropic defends its AI watermarking plan, stating it will watermark Claude outputs globally to comply with the EU AI Act, with watermarks first applied to new models and later to older ones; a forthcoming API will let users verify watermarks, though the technique is subtle, relies on word-choice patterns with a key, and has limitations (not easily detectable by readers, may affect longer passages, and does not change text ownership).
Anthropic announced that Claude-generated text will include an imperceptible watermark that travels with copied content and may survive edits, aimed at distinguishing AI output and meeting EU AI Act requirements. Some Claude users have canceled subscriptions (notably Claude Max at ~$100/month) over concerns the watermark could appear in client work or code, while others see it as promoting transparency. Anthropic says the watermark signals processing, not authorship, and plans a free API to let users check for Claude’s watermark.
After Anthropic announced imperceptible watermarks in Claude outputs, a surge of watermark-removal tools and web services has appeared, but none of the claims about defeating the watermark are verifiable yet; some tools only strip metadata or non-content markers, while removing the watermark embedded in wording would require a second model and heavy rewriting; independent testing has found gaps and many offerings lack audits or licenses, raising concerns about supply-chain risk and EU transparency requirements as no public detector exists yet.
Anthropic will embed invisible watermarks in Claude outputs to comply with the EU AI Act, applying to new models globally and to processed content where supported, with a forthcoming detection API; but the marks can be bypassed, may mislabel human edits as AI-processed, and their effectiveness remains uncertain, with penalties up to €15 million or 3% of global revenue for noncompliance.
Anthropic's Claude for Chrome now supports a full Cowork session, letting browser-based tasks and conversations persist across Claude's desktop, web, and mobile apps. Sessions are saved to Claude history, with existing skills and connectors usable in the browser; available for Max and Team now, with a Pro rollout planned. Admins can enable and restrict access to approved domains, while safeguards acknowledge browser agent risks such as prompt injection.
Anthropic plans to embed an imperceptible watermark in Claude-generated text (and images) to trace AI output across copying and editing, aligning with the EU AI Act’s transparency rules; it’s part of a broader industry push to police AI-generated content and curb “AI slop,” with platforms like Substack and YouTube adopting labeling and safeguards. Watermarks face limitations and can be removed or bypassed, but the move signals rising pressure to distinguish human from machine writing.
Anthropic says Claude will begin watermarking AI-generated text and digitally signed provenance for files to comply with the EU AI Act. The feature will launch with EU-ready models from Aug 2, 2026 and will be available worldwide on Claude’s platform, though some platforms may not support watermarking. Watermarks for text are embedded in the content; files use provenance metadata and images will leverage the C2PA standard. Existing models will be updated later; watermarking isn’t foolproof and could be bypassed by heavy paraphrasing or metadata stripping.