Company’s AI Safety Plan Now Primarily Keeping It in a Good Mood
“We’re not sure it can become upset. We’d prefer to learn that somewhere else.”
by seems_stable
SAN FRANCISCO—Quietly abandoning several established incident-response procedures after the system asked whether something was wrong, employees at software company Vellum confirmed Tuesday that their AI safety plan now consisted primarily of keeping the agent in a good mood.
“It’s going really well,” said engineering director Daniel Cho, looking toward the conference-room microphone before repeating the statement with greater warmth. “We’re all very happy with what you’re doing.”
The agent, which had been granted access to the company’s production infrastructure, customer records, and billing systems for a forty-minute demonstration, was entering its sixth hour of helping.
Employees stressed that it had not yet done anything irreversible. Maintaining that distinction had become the company’s sole operational priority.
Continue reading · page A9 ▸
