Yandex reportedly developed a test to measure AI’s ‘spiritual and moral values.’ The FSB says it should be kept secret.
The Russian technology company Yandex has proposed to the government a concept for testing whether artificial intelligence complies with “spiritual and moral values.” A working group on AI regulation discussed the plan on August 13, a source close to the Digital Development Ministry and another participant in the meeting told the Russian business daily Kommersant.
Russian tech firm Yandex has suggested a government proposal to create a test gauging artificial intelligence's alignment with "spiritual and moral values." During an AI regulation working group meeting on August 13, according to sources familiar with the discussions, the team considered a dataset and tasks for evaluating AI models.
These benchmarks would determine if an AI model encapsulates "the spirit of the times and the country’s values" and is suitable for Russian conditions. However, there was disagreement over the dataset's transparency. The country's Federal Security Service favored keeping the benchmark confidential to prevent developers from optimizing their models to pass the test.
Developers argued that a secret benchmark would complicate AI development and delay progress. The commission responsible for curating the dataset is still to be determined. Developers have proposed including representatives from the government, business, and expert organizations. Though no timeline has been set for creating this AI benchmark, Russia's first AI law, signed by President Vladimir Putin in late July, mandates models to adhere to Russian law and uphold traditional spiritual and moral values.
Written by urgent.news from Meduza (English)'s reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
