Who is responsible if the AI agent makes an incorrect trade?
"An incorrect trade" can mean two very different things, and they're treated differently. Every trade the agent places runs under your account, using the specific permissions you granted it. If the agent does exactly what you instructed but the trade itself performs badly — price moves against you, a strategy underperforms — that is ordinary investment risk, no different from a manual trade going wrong, and does not point to any fault in the agent connection: this is your risk to bear, the same as any manual order. If the agent instead misinterprets your instruction or executes something you did not actually intend, that is a genuine misexecution — for this scenario, raise it with Pluang support as soon as you notice it, and it will be reviewed case by case. Pluang's formal liability policy for these cases hasn't been published yet.
Distinguishing these two scenarios matters because they call for different responses:
- Scenario 1 — the agent did exactly what you asked, and the trade lost money. This is standard market risk. Every trade, manual or agent-placed, can move against you. Pluang does not guarantee returns on any trade, and an AI agent following your instructions correctly but landing on a losing outcome is not treated as an error — it's the same risk you'd accept placing the order yourself.
- Scenario 2 — the agent executed something you didn't ask for, or misread your instruction. This is a genuine misexecution, not routine market risk. If you believe this happened, see "What support is available if an AI agent executes a trade I didn't intend?" — that article covers the practical next steps for reporting it. Pluang's formal liability policy for these cases hasn't been published yet, and each case is currently reviewed by support individually.
- What doesn't change either way: the trade is still placed under your account, using the access and permissions you set up.
- Reducing the risk upfront: you can scope how much authority the agent has before anything goes wrong — see "Are there trading limits (per-transaction or daily) when using an AI agent?", which is a proactive way to cap potential downside from either scenario.
Related questions:
Q: Is a losing trade placed by my AI agent treated as an error?
Not on its own. If the agent carried out your instruction correctly and the trade simply lost money, that's normal investment risk — the same as any manual trade that doesn't work out. Pluang does not treat a correctly-executed trade with a bad outcome as an agent fault, and it does not guarantee returns on trades placed through Pluang Agentic Trading, whether manual or agent-initiated.
Q: What should I do if I think the AI agent executed a trade I didn't intend?
Treat it as a genuine misexecution rather than ordinary risk, and raise it through Pluang's support channel as soon as you notice it. See "What support is available if an AI agent executes a trade I didn't intend?" for the specific process.
Q: Can I limit how much my AI agent is allowed to trade to reduce this risk?
Yes. You can set per-transaction and daily trading limits before connecting the agent, which caps how much a single misunderstood instruction — or a single bad trade — can affect your account.
Q: Does Pluang guarantee compensation if my AI agent's trade loses money?
No. Pluang does not promise guaranteed remediation or compensation for trades that lose value, whether placed manually or through an AI agent — that would be ordinary investment risk.