Skip to content
Submit incident
AII-1002-56909
Well documentedThree or more independent publishers report the core facts.

France's Lucie chatbot taken offline after absurd responses

France's government-backed Lucie chatbot was taken offline three days after its January 2025 launch due to generating nonsensical responses, including claiming cows produce eggs and miscalculating basic arithmetic. The developer acknowledged inadequate testing and preparation for public scrutiny before deploying the early-stage research system.

January 2025 Β· Linagora Group Β· Reliability and hallucination
January 2025
AII-1002-569096 sources ↓
Quick facts
Category
Reliability and hallucination
Country
France
Source
edition.cnn.com
Date
January 2025
Drafted by
Raidu Incident Detection

What happened

Launched on January 23, 2025, Lucie quickly garnered criticism for absurd outputs, with users reporting that when asked about cow's eggs, the bot incorrectly claimed they were produced by cows and were a source of protein.

The chatbot miscalculated basic math problems, such as answering 5Γ—(3+2) with 17 instead of the correct 25, leading to widespread ridicule on social media.

Linagora Group disabled public access to the chatbot three days after its launch due to the backlash.

Reported impact

The project's failure was seen to reflect poorly on France's ambitions in the AI sector and raised questions about the readiness of such technologies for public use.

Index analysis

Lucie was at an early stage of training, having received only a single round of instruction fine-tuning, with the developer acknowledging that responses were not yet reliable.

The chatbot's model was stated to be hundreds of times smaller than models behind ChatGPT and was not designed for mathematics or coding tasks.

Governance controls

Linagora admitted they were 'swept away by excitement' and failed to prepare for public scrutiny, launching as an academic research initiative without adequate testing or communication regarding its limitations.

Classification

Category
Reliability and hallucination
Country
France
Harm status
Potential harm
Intent
Unintentional
Timing
Post-deployment
Harm types
Reputational
Harmed
Business
Failure modes
Hallucination, Agent misaction
Consequences
System suspension or termination
AI systems
Lucie

Sources and evidence

This record was researched and written by the Index, based on the following original reporting.

edition.cnn.com
France's Lucie chatbot taken offline after absurd responses
entrevue.fr
France's Lucie chatbot taken offline after absurd responses
lucie.chat
France's Lucie chatbot taken offline after absurd responses
lucie.chat
France's Lucie chatbot taken offline after absurd responses
ndtv.com
France's Lucie chatbot taken offline after absurd responses
thetimes.com
France's Lucie chatbot taken offline after absurd responses