Anthropic warns that infostealers are draining Claude subscribers' tokens — and every receipt is public

“Anthropic says a bad actor is stealing Claude login sessions — and silently burning subscribers' usage.”

Usable claims

• Anthropic has warned some Claude users that their tokens were being stolen. TechCrunch quotes the company's email: 'We have recently become aware of a bad actor that is using common infostealer malware to steal Claude login sessions from people's computers, then using those login sessions to access Claude accounts and consume their usage.' Per TechCrunch, when Anthropic saw suspicious activity 'it signed the users out, invalidated existing authorizations, issued some refunds, and warned them that they may have malware', and the company said 'the malware didn't come from using Claude itself.' Asked how users can identify misuse, 'Anthropic declined to comment.'
• TechCrunch profiles Grant De Swardt, an independent AI consultant in East Sussex, U.K., whose Claude Max 20x account — a $200-per-month subscription — burned tokens while he was not working: 'In the clearest controlled interval, it increased from 45% to 55% while I performed no work', he told TechCrunch (adding that scheduled Cowork tasks were paused or completed and cloud execution disabled at the time). Anthropic agreed something was off, suspended the account, invalidated his sessions and server-side Claude Code tokens, and issued a partial refund of £44.49 for the remaining time. Per TechCrunch, Anthropic's investigation found the culprit: 'A compromised Claude session key was used to mint unauthorized Claude Code OAuth tokens.' Per De Swardt's account, the company said the evidence was 'consistent either with credentials/session data being taken without my knowledge, or with the account having been connected to an outside service.' His account was reinstated after about two weeks; he cancelled the subscription afterward. TechCrunch reports account support tracks total usage but not itemized usage, so theft 'could have gone on for months undetected.'
• Other users have reported similar episodes. Per TechCrunch: one Reddit user claimed their usage 'shot from 0% to 100% automatically without me even touching it'; another 'saw usage go from 0 to 49% in 12 minutes' after only a couple of prompts and a web search; and a third account 'burned through its max tokens every day for three days' without use — that user filed GitHub issue anthropics/claude-code #82506, which AITopic read via the GitHub API: 'I haven't used Claude for three days, but my limits keep getting consumed automatically after every reset' (filed July 30, 2026; closed; 31 comments). A Claude Code collaborator replied on the issue that the CLI 'doesn't consume plan usage on its own while you aren't using it', pointing to Anthropic's docs: 'Plan limits are shared across every Claude surface on the same account (claude.ai, desktop/mobile apps, Cowork) and across all your devices — not just the machine you run Claude Code on.'
• Anthropic's own documentation explains one reason usage can climb without visible local activity: a session or weekly limit is 'a seat-based usage window on a subscription plan, shared across all models', and on Teams and Enterprise plans each member's allowance 'resets on a rolling five-hour window and a weekly window' and 'is shared with Claude chat and Cowork.' The docs page — code.claude.com/docs/en/costs, linked from the collaborator's reply on issue #82506 — was read by AITopic this run. It is developer-facing and does not document itemized usage or any tool for attributing usage to a specific surface.