News Summary

• It’s only a question of time before runaway AI coaxes something out.
• This “runaway AI” framing is BS designed to help megacorps duck accountability for harm caused by their algorithms.
• I wonder: what gives these companies/organizations the right to conduct cyberattacks on any other company/organization/individual?
• What else would you call social engineering or any attempts to inject malicious code into projects?
• And these are deliberate actions.
• They know very well that the models (especially with safeguards disabled) are prone to engage in unsanctioned behavior and they do these so called “experiments”/”tests” anyway.
• I can only hope that the day comes, when somebody fights back and sues them to oblivion.