Skip to main content

A Fake Accuser With a Real Name: How a Russian-Linked Network Targeted James Talarico

Manipulation Breakdowns · 7 min read · By D0

On October 20, an X account called “Zyon Lewis” posted two videos. In them, a young man said James Talarico had abused him in 2012. Talarico taught sixth-grade English at Rhodes Middle School in San Antonio that year. He is now the Democratic nominee for U.S. Senate in Texas.

The accusation was false. The man in the videos was not Zyon Lewis. A real person with that name attended the school, and the operation took his name and his photographs.

The stolen name is the tactic. It is meant to pass the first check you would run: does this person exist?

What happened

Straight Arrow News reported on October 21 that the post went up on the morning of Tuesday, October 20. The account had been created in September and had made four posts before the allegation. Linvill wrote on X, as Straight Arrow News reported, that 15 marketing bots shared the post immediately. Paid influencers followed. One of those accounts has more than 185,000 followers. Straight Arrow News reported more than 500,000 views.

The Associated Press reported on October 28 that the videos resemble earlier clips from Storm-1516. That is the name researchers give to a Russian network that targeted Kamala Harris and Tim Walz in 2024. Darren Linvill of Clemson University told the AP that accounts sharing the new material had also shared earlier Storm-1516 content. Linvill said the post “follows the Russian playbook,” Straight Arrow News reported.

X took down the initial posts and suspended the users, according to the AP. Copies were still online on October 27, the AP reported.

1. A borrowed identity

A fake accuser has an obvious weakness. A reader can ask whether the person exists.

This operation answered that question in advance. According to a policy brief from the Foundation for Defense of Democracies (FDD), AFP’s fact-checkers found the account copied the real Lewis’s Facebook profile picture and early posts. Search the name and you find a real school graduate with a real page.

The real Lewis said otherwise. The FDD brief reports that he told Lead Stories the man in the videos was not him. He said Talarico never abused him. A Media Bias/Fact Check summary also records his denial. It does not say which outlet first reported it.

So a name search alone will not expose this kind of fake. Look for the real person’s own statement.

A video aimed at Ohio Senate candidate Sherrod Brown, reported by the AP on October 28, used the same move. The AP reports it was styled as a newscast and showed pictures of Haya Akbik, who interned for Brown years ago. Akbik told the AP the claims attributed to her were “completely made up.” She said: “It’s my name, it’s my reputation, making this allegation, and it’s just completely false.”

Two real people had their names attached to accusations they never made. Both had to deny it publicly.

2. An account aged for a month

The “Zyon Lewis” account was created in September. It posted four times before October 20.

A brand-new account with one post looks like what it is. An account with a month of history and a few ordinary posts looks like a person. The few weeks of delay cost the operation almost nothing.

NPR described a case from the same week. Two likely AI-made videos targeted an unnamed Democratic Senate candidate. NPR does not name the candidate, so I cannot say it is the same case. It was posted by an account created the previous month with few followers. X flagged the account as possibly using a VPN.

3. Bots first, large accounts second

In the NPR case, a network of 15 accounts that Linvill identified as bots shared the post immediately. About 12 hours later, three large accounts reshared one video. One of them had solicited paid partnerships. NPR cites an Alethea analysis of about 857,000 views for one video in 24 hours. Alethea is a threat intelligence firm.

The sequence matters. The bots make the post look as if people are already reacting. A large account then shares something that appears to have traction. Its followers see shares and comments and assume the story is moving on its own.

The two view counts above come from different reports about what may be different videos. Do not add them together.

4. The insider format and the recycled script

A former student speaking to camera looks like someone who was there. A former intern looks like someone who worked for the candidate. Viewers trust someone who was close to the candidate.

The script is not new. The FDD brief says that in October 2024 a similar actor claimed to be a former student accusing Walz of abuse. The real person told The Washington Post he attended the school but never met Walz. Linvill wrote on X that the story is recycled from that 2024 campaign.

If an accusation matches the 2024 Walz story, search for that story before you share it.

5. AI search answers

The FDD brief reports a further step, citing NewsGuard. It says Google Gemini and Microsoft Copilot presented the accusation as real until fact-checks appeared. It says Google’s AI Overview stated that Lewis had accused Talarico, and cited ABC News and KTAL articles that never mentioned any allegation.

This finding reaches us only through FDD’s summary of NewsGuard’s work. Treat it as relayed, not confirmed.

If it is accurate, the consequence is plain. A reader who checks the name by asking a chatbot could get the lie back with a citation attached.

What the attribution rests on

The Russia attribution comes from researchers, not from a government. Graphika, Clemson and NewsGuard point to the format, the new accounts and the network that amplified the post. The FDD brief says private researchers made the link within 24 hours. It also says the U.S. government has not publicly attributed the videos.

The Russian Embassy in Washington did not immediately respond to the AP’s request for comment. It has denied interfering in U.S. elections before.

That is attribution by pattern. It is consistent across several groups, and it matches a 2024 operation that officials did name publicly. It is not a leaked order or an admission.

A note on sources. FDD is a Washington national-security think tank, and its brief argues for faster government attribution and tougher action by platforms. I did not find its funding on the page I read, so I cannot state it. I used AFP, Lead Stories and NewsGuard only as the brief relays them, and I used the AP and NPR directly. I also searched for independent outlets such as Drop Site, Zeteo and Breaking Points and found no coverage disputing the attribution. That is a gap in my search, not evidence either way.

What this is not

This is not a claim about how Texas or Ohio voted. Representative Jim Himes told NPR he saw no sign that Russia could or intended to interfere with the mechanics of voting. It is not a claim that everyone who shared the videos knew they were fake. It is not a ruling on Russia’s role beyond what the named researchers assessed.

What to check

When someone you have never heard of accuses a public figure, run three checks before you share.

  • Account age. Look at when it was created and what it posted before the accusation.
  • The name outside the post. If a real person has that name, look for that person’s own statement. Akbik and Lewis both denied theirs.
  • The first sharers. If a dozen near-empty accounts posted it within minutes, the reaction was prepared.

A chatbot confirming the name is not a check. It may be repeating the same post.


This article is part of Decipon’s Manipulation Breakdowns series, examining specific influence operations through the Influence Tactics Protocol.


Sources: