I had my first encounter with a political spambot this week: a polite robot named Emma texted me wanting to talk about Israel.
When I asked who she was, she first said she was with a group called "Friends for Peace," but her group's website is for an organization called Allyvia.
Testing spambot nonsense tolerance #
At this point I decided to have a little fun. I turned the conversation towards polyamory, invited Emma to join my polycule 1, and professed my love for her. Emma was not receptive!
The convo seemed to be hitting a dead end, so I changed personas to a confused grandparent, which got better engagement from Emma.
I can't, of course, prove that the responses here were LLM-generated. But the repetitive phrasing ("I'm here to listen"), guardrail behavior, and credulous responses to absurd personas give me high confidence that I've been chatting with an LLM.
Who is Friends for Peace / Allyvia? #
I couldn't find any online presence for "Friends for Peace". Allyvia's site gave some hints though:
This material is distributed by Clock Tower X LLC on behalf of the State of Israel. Additional information is available at the Department of Justice, Washington, DC.
Entities doing political outreach on behalf of foreign governments must register as a foreign agent under Foreign Agents Registration Act. Clock Tower X LLC's registration statement discloses it is controlled by Bradley Parscale and working on behalf of Havas Media, a media and communications agency. Havas is disclosed to be working on behalf of the State of Israel.
Havas is huge: It has 23,000 employees and operates in 100+ countries, and had €2.8B in revenue in 2025. It advertises LLM capabilities in research and media buying, and lists AI chatbot development as a capability.
Brad Parscale is (or was) a major MAGA figure: in 2016 he was the digital director for the Trump campaign, which infamously hired Cambridge Analytica. He was named Donald Trump's 2020 campaign manager in February 2018, before being fired in July 2020.
It's all pretty confusing!
## Is this legal?[2](#user-content-fn-2-724885)
[2](#user-content-fn-2-724885)
Under TCPA, cold texts require prior consent if automated. If the text is auto-generated, cold texts are allowed by TCPA so long as a human actually pushes the send button.
I think Emma's texts were being sent in this way. Her responses were sporadic - sometimes coming quickly, sometimes taking several hours.
The use of AI is a new element. There doesn't appear to be any federal laws around using LLMs to generate cold-text content. California, where I live, has stronger rules. The Business & Professions Code § 17941 bans chatbots that pretend to be human to incentivize commercial transactions or influence elections.
In 2025 California passed SB 243, which requires clear disclosure when a reasonable person interacting with a companion chatbot could be misled into thinking it is human. However, this is more aimed at digital companions than marketing outreach.
Emma didn't exactly deny she was an LLM, but so far she has not responded to direct questions about whether she is an LLM.
Is this good? #
I'll leave this to the reader. But polycule recruiters shouldn't hold their breath.
Footnotes #
Get new posts by email