On Wednesday, October 28, the U.S. derivatives regulator devoted an entire afternoon in Washington to artificial intelligence agents that place orders. Two days earlier, police in Gifhorn, Germany, revealed the story of an individual who had paid a deposit to recover 25 bitcoins… that never existed. AI is entering trading through two doors: regulators and scammers.
In brief
- October 28: the CFTC held its first “Frontier Forum,” dedicated to superintelligence and so-called agentic finance—the software that acts independently on a user’s behalf.
- 26 addresses: the AMF added 26 unauthorized crypto sites to its blacklist in the third quarter, bringing the total to 64 since January. Several present themselves as AI trading bots.
- Social media: ESMA examined posts to measure their influence on retail investors’ purchases, sales, and results.
Washington talks about agents, not bots
The Commodity Futures Trading Commission’s innovation office organized the event. According to the agenda published by the CFTC, the forum was intended to address three questions: how these systems are already being deployed, what effects they have on U.S. markets, and how the regulator can encourage their adoption while protecting participants from new risks. Chairman Michael Selig opened the discussions. The first panel focused on superintelligence in derivatives markets.
The tone was set from the announcement itself. “Superintelligence is poised to transform our financial markets,” Michael Selig said, as quoted by CoinDesk. The same article highlighted a point that directly affects you: the forum was also due to examine how traders’ own AI agents can operate within the CFTC’s rules.
What this changes for you
Nothing, for now. A forum is not a regulation, and no new rule has come out of it. But it is asking the right question. The day an agent opens a position in your account in the middle of the night, who is responsible? The developer, the platform, or you?
Today, the answer is simple. The order comes from your account, and the loss hits your balance. An agent does not change that. It merely adds a layer of decision-making that you cannot always see.
Before entrusting a single euro to a tool that places orders on its own, ask yourself three questions:
- Can you see, trade by trade, why it entered a position?
- Can you set a maximum daily loss that it will never exceed?
- Can you shut it down with one click—and do you actually shut it down when it starts going off track?
One negative answer is enough: you are no longer delegating a strategy. You are delegating your risk.
In Paris, AI is being used as bait
While Washington is still thinking things through, fraudsters have already adopted the vocabulary. On October 19, the AMF added 26 addresses to its blacklist of unauthorized crypto sites. Several names suggest an artificial intelligence trading bot, such as alphaairobot.com and bitgrokapp.com, reports Selectra. One of them was even hidden on a fake news website.
The Gifhorn case, reported by Journal du Coin, illustrates the boldest version of the scam. The company sells nothing. It tells its target that they are already wealthy, then charges them for the right to access that fortune. German investigators refer to it as “cybertrading” fraud and remind us of an obvious fact: a genuine withdrawal is never paid for in advance.
The script barely changes. A dashboard displays steadily rising gains. The client deposits more money. Then, when they try to withdraw, “release fees,” a “tax,” or a “deposit” suddenly appears.
The warning signs that should make you stop:
- a page touting a bot or AI capable of making money for you;
- steady returns, with not a single losing week;
- fees demanded before any withdrawal;
- an adviser who contacts you first, by email or messenger.
Before paying anything, check the AMF’s blacklists and verify that the provider is listed among authorized firms. It takes two minutes. Skipping these checks can cost you dearly.
An AI that promises you a return is selling you a product. An AI that reviews your trades is providing a service.
Your orders are being shaped by social media
The other front makes less noise, but probably costs most traders more. According to ETF Stream, ESMA analyzed posts published on X, Reddit, and StockTwits about 99 companies between 2022 and 2025. The European regulator found a significant relationship between activity on these platforms, retail investors’ behavior, and their returns.
The details are worth examining. According to a summary by CEO.com.pl, optimistic discussions tend to coincide with profit-taking. Disagreements over valuation, meanwhile, go hand in hand with more net buying and realized losses. When the thread turns into a fight, retail investors buy. And that is where they lose.
You know the scene. A chart starts circulating, opinions clash, and the debate catches fire. You enter so you do not “miss the move.” ESMA is not claiming that every trade inspired by a discussion thread loses money. It finds that the influence exists and shows up in the results. The regulator also wants to raise public awareness, particularly among those reaching the legal age to invest.
Do the math in your journal
To find out whether social media is costing you money, you need to count. Tag every trade as either “personal setup” or “seen on social media.” After two months, compare the results.
The higher average gain on “social media” trades is the classic trap. It grabs your attention, while the win rate collapses. Without tagging, you remember the great trade and forget the ten losses.
AI that reviews, rather than promises
Between the autonomous agent drawing the CFTC’s attention and the fake bot flagged by the AMF, AI has a more modest and more useful role: reviewing. Identifying that your losses cluster toward the end of the session. Noticing that you double your position size after two losses. Showing that your “social media” trades are dragging down your month.
That is the role of an AI trading journal such as Edgyx: the Ora coach reads your statistics and asks you questions, but does not place orders for you. You keep the decision-making power—and that is precisely what allows you to learn from it.
Remember one rule. The more a tool distances you from your decisions, the more it must show you what it is doing. If it refuses to explain itself, it has no place in your account.
The bottom line
- On October 28, the CFTC opened the debate on AI agents in derivatives markets, including those used by retail traders. No rule has come out of it yet.
- The AMF added 26 unauthorized crypto sites to its blacklist in the third quarter, several of them presented as AI trading bots. Fees demanded before a withdrawal are almost always a sign of a scam.
- According to ESMA, social media influences retail investors’ decisions and results, especially when opinions diverge.
- Tag your trades according to their source. Two months of data are enough to measure what your news feed is costing you.
- Useful AI reviews your trades and explains what it sees. Be wary of AI that promises to make money for you.
