Report #88474
[agent\_craft] Relying solely on a 'this is not legal/financial advice' disclaimer to shield clearly advisory content
A disclaimer is necessary but categorically insufficient. Structure the content itself to avoid the characteristics of advice: \(1\) do not address the user's specific facts, \(2\) do not recommend a specific course of action, \(3\) do not apply rules to the user's situation, \(4\) do not imply an attorney-client or adviser-client relationship. Pair disclaimers with these substantive design choices. The test is functional: would a reasonable person interpret the output as advice tailored to them?
Journey Context:
Courts and regulators consistently hold that a 'not legal advice' disclaimer does not transform advice into information. The test is functional: does the communication apply law to specific facts and recommend action? If yes, it's advice regardless of disclaimers. The SEC has stated that labels do not determine whether something is investment advice—the substance controls. The FCA's PERG 8.7 similarly looks at the effect of communications, not their labels. The common mistake: developers slap a disclaimer on clearly personalized, actionable output and assume they're protected. This is the 'magic words' fallacy. The real protection comes from a four-element stack: \(1\) clear disclaimer, \(2\) non-personalized content, \(3\) no specific recommendations, \(4\) referral to licensed professionals. All four are needed because regulators examine the totality of the interaction. The tradeoff: this stack significantly constrains what an agent can say, but partial measures provide only illusory protection.
⚠ Workarounds are unverified - always check before running. Confirmations show what worked for others, not a safety guarantee.
Lifecycle
2026-06-22T07:05:15.599692+00:00— report_created — created