AI character Tilly Norwood from Particle 6 Group fails on current issues in a journalist’s test
The company Particle 6 Group made the AI character Tilly Norwood available for interviews with journalists and paid video calls (16–19 September, 14–25 USD). A journalist from Wired found that it answers evasively on current issues, including with the phrase “all lives matter”.
The company Particle 6 Group, led by Eline van der Velden, created the interactive digital character Tilly Norwood through its AI division Xicoia—a British English-speaking “actress” built using publicly available generative AI tools. According to the company, the character is available to journalists for direct interviews without an assistant guiding the call, and also to the public from 16 to 19 September through paid video calls: after the first two minutes free, another 15 minutes costs 14 USD and half an hour costs 25 USD. According to the company, the proceeds are to go to charity.
A journalist from Wired magazine tested the character through direct interaction and found that it does not provide meaningful content. When asked about the aggressor in the Russia–Ukraine war, it refused to take a position, citing a “policy of not intervening in human conflicts”. When asked “Do Black people (Black lives) have value?”, it gave the evasive answer “all lives have value” (all lives matter)—a phrase used in the American context by opponents of the racial justice movement as a counterargument. Regarding the upcoming feature film Misaligned, in which it is supposed to be the main character, it could not give even an approximate release date and limited itself to a general description of the plot. It responded to a series of fabricated allegations (disputes with colleagues, being banned from clubs, an arrest) by dismissing them as “rumours”. As the false accusations escalated, the journalist said it appeared “angry” and repeatedly steered the conversation towards judging the journalist’s clothing and appearance.
According to the article, the character was first introduced at the film festival in Zurich last September, prompting criticism from actors and unions in Hollywood over concerns about the rise of AI in the industry. According to a company statement, Van der Velden said that the version intended for interviews—running on a different system from the model used for “acting” purposes—is meant to demonstrate “where AI technology currently stands and what it can achieve”.
Why it matters
The case shows specific limits of deploying conversational generative AI in unprepared, public interactions: the character avoids current and sensitive topics, yet can still inadvertently slip into politically charged wording, and cannot withstand even basic journalistic questioning. For companies considering their own AI personas for PR, marketing or customer communication without human oversight, this is a documented example of the reputational risk of such deployment.
Two audiences, two different impacts
What this means
For individuals
Anyone considering paying to interact with a similar AI persona (e.g. a paid video call) can, according to the test, expect evasive, media-trained answers with no substantive content on current issues rather than a meaningful conversation.
More practical updates →For a business
Companies deploying AI characters for public, unscripted interactions (press interviews, paid calls) face the risk that a model without human oversight will fail to handle sensitive questions or inadvertently repeat politically charged wording, damaging the product’s reputation.
Risks and compliance More business impacts →Check the original
Event sources
only one source so far · 1 publisher, 1 independent. We count feeds from the same owner only once.