“Anthropic CEO Dario Amodei called on other AI companies to conduct mandatory security testing of neural networks before their public release. Bloomberg writes about it. “I think we should make testing mandatory, but it’s also important to be very careful about how we do it,” he said. According to the head of an AI startup, there are a number of voluntary security rules that large firms have agreed to. However, clearer standards are needed. “There is nothing that would give […]”, — write: businessua.com.ua
Anthropic CEO Dario Amodei called on other AI companies to conduct mandatory security testing of neural networks before their public release. Bloomberg writes about it.
“I think that we should make testing mandatory, but it’s also important to be very careful about how we do it,” he said.
According to the head of an AI startup, there are a number of voluntary security rules that large firms have agreed to. However, clearer standards are needed.
“There is nothing to verify or verify compliance by the company. They just said they would do it,” he emphasized.
Amodei added that startups have moved to adhere to a set of standards influenced by the public and workers, but that ultimately won’t be enough. At the same time, testing requirements must remain “flexible” to take into account the pace of technological change.
In November, Anthropic opened its AI products to US intelligence and defense agencies. Earlier, Meta took a similar step.
It will be recalled that in the same month Amodei announced that artificial intelligence will reach human level in 2026.
The source