Description: Polish radio station Off Radio Krakow replaced human presenters with AI-generated ones and aired a simulated interview with the deceased poet Wisława Szymborska. The AI-driven experiment aimed to attract younger listeners but led to job losses for former hosts. In response to the public backlash, the station ended its use of AI presenters.
Editor Notes: Michal Rusinek, who operates the foundation that runs Wisława Szymborska's estate, reportedly gave permission for her voice to be used, but he later expressed disappointment with the outcome, describing the AI-generated interview as “horrible” and saying it made Szymborska sound "bland" and "naïve." Rusinek noted that, while Szymborska herself might have found humor in the experiment, the interview misrepresented her tone and personality. Regardless of Rusinek's role and response, Incident 833 still raises thorny problems about the ethical dimensions of using AI to recreate the voices of deceased individuals, along with all the potential and actual harms that result from eroding transparency and truth in media platforms. See also Incident 627: Unauthorized AI Impersonation of George Carlin Used in Comedy Special.
Entities
View all entitiesAlleged: OpenAI , ElevenLabs and Leonardo AI developed an AI system deployed by Off Radio Krakow and Mariusz Marcin Pulit, which harmed Lukasz Zaleski , Mateusz Demski , Wisława Szymborska , Off Radio Krakow audience and Off Radio Krakow employees.
Incident Stats
Risk Subdomain
A further 23 subdomains create an accessible and understandable classification of hazards and harms associated with AI
6.2. Increased inequality and decline in employment quality
Risk Domain
The Domain Taxonomy of AI Risks classifies risks into seven AI risk domains: (1) Discrimination & toxicity, (2) Privacy & security, (3) Misinformation, (4) Malicious actors & misuse, (5) Human-computer interaction, (6) Socioeconomic & environmental harms, and (7) AI system safety, failures & limitations.
- Socioeconomic & Environmental Harms
Entity
Which, if any, entity is presented as the main cause of the risk
Human
Timing
The stage in the AI lifecycle at which the risk is presented as occurring
Post-deployment
Intent
Whether the risk is presented as occurring as an expected or unexpected outcome from pursuing a goal
Intentional
Incident Reports
Reports Timeline
WARSAW, Poland -- A Polish radio station said Monday that it has ended an “experiment” that involved using AI-generated "presenters" instead of real journalists after the move sparked an outcry.
Weeks after dismissing its journalists, OFF R…
When a state-funded Polish radio station canceled a weekly show featuring interviews with theater directors and writers, the host of the program went quietly, resigned to media industry realities of cost-cutting and shifting tastes away fro…
Variants
A "variant" is an incident that shares the same causative factors, produces similar harms, and involves the same intelligent systems as a known AI incident. Rather than index variants as entirely separate incidents, we list variations of incidents under the first similar incident submitted to the database. Unlike other submission types to the incident database, variants are not required to have reporting in evidence external to the Incident Database. Learn more from the research paper.
Similar Incidents
Selected by our editors
Did our AI mess up? Flag the unrelated incidents
Robot kills worker at German Volkswagen plant
· 27 reports
Nuclear False Alarm
· 27 reports
Amazon Censors Gay Books
· 24 reports
Similar Incidents
Selected by our editors
Did our AI mess up? Flag the unrelated incidents
Robot kills worker at German Volkswagen plant
· 27 reports
Nuclear False Alarm
· 27 reports
Amazon Censors Gay Books
· 24 reports