Probably, in our judgement, though there is no reliable public data yet on agents buying into vetted token raises. Bots already trade new tokens on open launches, and one study of a popular launchpad counted 2.6 million trading addresses in a single month. Raises built to check people will need to check what stands behind each agent, too.
A note on the legal and regulatory points here: they summarise public sources as of November 2026 and should not be relied on. Rules for AI agents are still being written, and some of what is described here is proposed or coming soon rather than settled. Check the current position, or ask a qualified professional, before acting on it.
Are AI agents already buying into token launches?
Automated buyers already trade on open token launches; what is not yet documented is agents investing in structured, vetted raises. Those are two different things, and the difference matters for everything below.
The open-launch side is well measured. An academic study of a popular memecoin launchpad covered September 2025. It counted 655,770 new tokens, 243,123 creator addresses and 2,600,790 trader addresses. The authors note that the true number of people is likely lower, because one person can run many wallets.
The same study found that markets dominated by bot-like trading had lower odds of success beyond the middle stages. Launch tools have responded to automated buyers. One token launcher’s agent skill offers an “anti-sniper” fee that can start as high as 80% and fall away over up to two minutes. A sniper is a bot that buys a token in its first moments.
What we could not find is solid data on agents committing money to vetted raises, with allocations, checks and lockups. That is the honest state of the evidence as of November 2026. Anything firmer would be a guess.
How is a token raise different from a launch an agent can buy into?
A token raise is a gated sale: buyers are checked, allocations are capped, and tokens are often locked up. An open launch lets any wallet buy, instantly, with no questions. A raise, in this guide, means a sale of tokens to early backers before or at launch.
| Open launch | Vetted token raise | |
|---|---|---|
| Who can buy | Any wallet with funds | Buyers who pass the sale’s checks |
| Identity check | None | Often a know-your-customer (KYC) check on a person or company |
| Allocation | Whatever you can buy, at whatever price | A capped amount per participant |
| Timing | First seconds favour bots | A set window, often with a lottery or queue |
| After the sale | Tokens tradable at once | Tokens often released on a vesting schedule |
| Who can be asked about it | Often no one | The project and the raise platform |
That table explains why agents fit open launches so easily and vetted raises so awkwardly. Every feature in the right-hand column assumes a person or company on the other side.
None of this stops an agent from taking part. A person who passes the checks can still hand the buying to an agent. The raise then knows who the investor is, but not what software is making the decisions, or under what limits. That gap is the subject of the next question.
Can an AI agent pass a KYC check?
Not as itself. KYC checks are built to verify people and companies, so any check on an agent falls on whoever stands behind it. The open question is whether that is enough.
Regulators are asking. At a CFTC forum on 28 October 2026, the CFTC chair noted that the agency’s rules were built around identifiable market participants. He asked whether a new kind of check, “know-your-agent”, might be needed, and whether an agent’s developer would become subject to registration. He said the views were his own. The remarks, listed in the CFTC’s index of public statements and remarks, pose questions rather than set rules.
US lawmakers are asking too. On 23 June 2026, eight Democrats on the House Financial Services Committee wrote to the SEC about AI agents that trade for retail investors. They asked about the responsibilities of brokers and of AI developers, and whether current securities law is enough. They requested a reply by 31 July.
On-chain identity is not ready to fill the gap. A June 2026 study of the ERC-8004 agent-identity standard found that only 3% to 15% of registered agents, depending on the network, had a valid registration file with a live service endpoint. A raise cannot lean on that registry as proof of who is behind an agent.
What could go wrong?
The central risk is that a raise built to check people gets filled by software that is hard to tell apart. Several specific failures follow from that.
One operator, many agents. This is a “sybil” problem, named for one identity posing as many. Caps and lotteries work per participant. If one operator runs fifty agents with fifty wallets, a per-wallet cap stops meaning anything. The launchpad study’s note about one person running many wallets applies here directly.
Agents chasing yield are easy to pitch. An agent told to “find high returns” reads pitches, not people. The CFTC has warned that AI cannot predict markets and that promises of guaranteed or very high returns are a red flag. An agent with no such rule may not treat them as one.
Crowding. Our inference: agents built on similar models and fed similar data may favour the same raises at the same moment. That can oversubscribe a raise in seconds and shut out people who read the documents first.
Money locked longer than intended. Raise allocations often vest over months. An agent may commit funds without weighing that those funds cannot be sold or moved until the schedule allows.
No clear owner of the decision. If an agent invests someone else’s money in a raise that fails, it is not settled who answers for it. That is exactly the question the CFTC chair and the lawmakers’ letter raise.
The trust takeaway
AI agents will probably reach token raises, and that makes vetting and identity checks more important, not less. The check has to land on the person or company behind the agent, not on the wallet. What to check depends on which side you are on.
Neither list is exotic. They are the questions a careful investor and a careful raise already ask. Agents simply remove the pause in which a person would ask them.
If you are giving an agent money:
- Can your agent commit funds to a raise at all, or only trade within set limits?
- Does it understand lockups, and will it tell you before tying money up for months?
- Does it apply your rules on guaranteed-return pitches, or just chase the highest number?
If you are running or joining a raise:
- Are caps applied per verified person or company, or per wallet?
- Does the raise allow automated participation, and does it say so plainly?
- Are the vesting and lockup terms published in a form an agent can read and a person can check?
TrustSwap Launchpad runs vetted token raises. Where this is going: raises that can verify the identity and authority behind an agent’s wallet, and terms that an agent can read before it commits. That is a direction, not a shipped feature.