Sanoman droonimoka paljasti tekoälyn ongelmat. Journalismin ”pellin alla” vaikutukset voivat kuitenkin olla vielä isommat, sanoo Laura Saarikoski.
| Source: Mastodon | Original article
Sanoma’s recent “drone‑meme” story has become a cautionary tale for the media industry. On Sunday the newspaper published a report claiming that a swarm of unmanned aerial vehicles had been spotted over Kouvola, only to retract the piece hours later when editors discovered that the text had been generated by an internal AI tool that hallucinated the entire incident. Chief editor Erja Yläjärvi confirmed that the error stemmed from over‑reliance on the system, which had not been cross‑checked against any independent source.
The episode matters because it exposes a structural weakness in modern newsrooms: the temptation to let language models draft copy without rigorous human verification. While AI can accelerate reporting, it also amplifies the risk of “fabricated facts” slipping through editorial filters, eroding public trust in a sector that traditionally enjoys high credibility in the Nordics. Media scholar Laura Saarikoski warned that the fallout could be far larger than a single false drone story, arguing that AI‑driven shortcuts may gradually shift the “soil” of journalism, making it harder to distinguish fact from algorithmic speculation.
What to watch next is the response from Sanoma and the wider Finnish press. The publisher has launched an internal audit and promised tighter safeguards, including mandatory fact‑checking of any AI‑generated text before publication. Industry bodies are already discussing a unified code of conduct, and the Finnish Media Foundation is funding research into AI‑journalism interactions. Regulators are also keeping an eye on the issue, as the EU’s AI Act moves toward stricter transparency requirements for high‑risk AI applications. The next few weeks will reveal whether Sanoma’s misstep triggers concrete policy changes or remains an isolated lesson in the growing pains of AI‑augmented newsrooms.
Sources
Back to AIPULSEN