Daniela Amodei

Factual · AI safety and alignment

“Because we have the information about how the technology can be abused, because we see it every day.”

Daniela Amodei, Anthropic president · 29 June 2026 · Daniela Amodei Says Curiosity Is Underrated (Stanford GSB)

Technology companies have information about how AI can be abused because they see it daily, via safeguards and security teams.

Mostly true True in substance, wrong in a detail.

The second judge said resolved true; the verdict shown is the first judge's.

Said for an organisation they lead or work for; it counts on both records.

What would count as true or false

True: AI developers (such as Anthropic) operate safeguards, trust-and-safety or threat-intelligence teams that routinely, on a daily or ongoing basis, observe and document misuse of their AI systems, as shown in published threat-intelligence or misuse reports. Mostly true: such monitoring exists and is ongoing, but 'every day' is an overstatement of how often misuse is detected or reported. Mixed: monitoring exists only at some companies, or the visibility into misuse is limited. Mostly false: companies have only sporadic or minimal visibility into misuse. False: the evidence shows companies do not monitor or see misuse of their systems.

Evidence

1. Countering misuse of AI: September 2026 / Anthropic Anthropic Own record

   “Over the past eight months, our Threat Intelligence team identified and disrupted operations in which threat actors tried to use Claude for malicious activity.”

2. Countering misuse of AI: September 2026 / Anthropic Anthropic Own record

   “The cases we share here aren’t typical misuse, but rather examples of the most notable and novel threat activity we’ve identified to date.”

3. Building safeguards for Claude Anthropic · 12 August 2025 Own record

   “We also monitor harmful Claude traffic, going beyond single prompts and individual accounts, to understand the prevalence of particular harms and identify more sophisticated attack patterns.”

4. GTIG AI Threat Tracker: From Prompting to Autonomy – The Evolution of Adversarial AI | Google Cloud Blog Google Threat Intelligence Group (Google Cloud Blog) · 8 September 2026

   “We continuously harden our AI models against misuse by feeding insights from active threat monitoring directly into our safety classifiers and guardrails.”

Reasoning

Anthropic's published materials describe monitoring of harmful Claude traffic to understand the prevalence of particular harms and identify attack patterns [E3]. Its threat-intelligence report says the team identified and disrupted malicious operations over the preceding eight months (E1). Google's threat intelligence group, a source independent of the subject, separately describes active, continuous threat monitoring of its AI models (E4), so such monitoring exists at more than one developer. However, the reports present selected notable cases, not a typical or complete record (E2), and none of the evidence shows misuse being detected literally every day, so "every day" is not established.

Cite this claim

Daniela Amodei, “Technology companies have information about how AI can be abused because they see it daily, via safeguards and security teams.” (said 29 June 2026). Verdict: mostly true. Reliability API, claim c_042f3c63dfa1, method 0.1.0+28da144b. https://epistacklabs.com/claims/c_042f3c63dfa1