The X (formerly Twitter) account Balance ton Claude, which previously exposed author Thélyson Orélien, claims to have identified two additional writers who allegedly used artificial intelligence to draft their novels. Samuel Fitoussi of the watchdog group states that these names may soon be released to the press, highlighting growing concerns over AI-generated literature.
The Bottom Line
- The Core Claim: The X account Balance ton Claude asserts it has found two more authors who relied on artificial intelligence to write their novels.
- The Detection Tool: Members of the group rely heavily on the AI detection software Pangram, claiming its false positive rate is extremely low at roughly 1 in 20,000.
- Contradicting Results: Prominent literary figures have challenged the software’s accuracy, with Quebec author Bryan Perro testing his 2006 youth novel Amos Daragon and finding that the tool flagged 73% of it as AI-generated.
Expanding the Scope of AI Scrutiny in Publishing
The controversy surrounding the intersection of artificial intelligence and creative writing continues to widen. Following the initial public scrutiny directed at Thélyson Orélien for his book C’était ça ou rien—which a Pangram scan reportedly indicated was more than 90% generated by AI—the digital watchdog group Balance ton Claude is expanding its investigations. Samuel Fitoussi, speaking on Noovo Info, indicated that the group’s ongoing efforts aim to make readers aware that multiple contemporary texts may have been produced using tools like ChatGPT.
“Quand l’auteur génère son texte en moins de temps qu’il en faut au lecteur pour le lire, il y a un manque de respect,” Fitoussi asserted during his media appearance, emphasizing the ethical concerns behind rapid, automated manuscript production.
The Reliability Debate Surrounding Detection Software
At the center of this movement is Pangram, the detection software utilized by Balance ton Claude to evaluate suspected texts.
However, the software’s infallibility has faced immediate pushback from established creators. Quebec author Bryan Perro decided to put the program to the test using his well-known youth fantasy series, Amos Daragon. To his surprise, the detector claimed that 73 percent of the book—which was originally published back in 2006, long before the current generative AI boom—consisted of AI-created content.
| Book Title | Author | Publication Year | Pangram AI Detection Score |
|---|---|---|---|
| C’était ça ou rien | Thélyson Orélien | Not specified in source | Over 90% generated via AI (per group report) |
| Amos Daragon | Bryan Perro | 2006 | 73% flagged as AI-generated (per author test) |
The Broader Cultural Fallout and Next Steps
Despite questions raised by authors regarding the accuracy of detection tools when applied to older or conventional works, the watchdogs maintain that their campaign has achieved its primary goal. Fitoussi declared the mission accomplished in terms of raising public awareness, noting that “tout le monde” is now running scans on artistic works using Pangram.
“Notre mission c’était de sensibiliser pour le contenu IA, c’était notre devoir.”
