Urgent.News

What's breaking now, across thousands of outlets.

AI

Sam Altman: Society Should Accept Some ‘Bad Things’ to See AI Benefits

OpenAI CEO Sam Altman says society should accept some AI harm to preserve the benefits and broad accessibility of artificial intelligence. The post Sam Altman: Society Should Accept Some ‘Bad Things’ to See AI Benefits appeared first on TechRepublic .

OpenAI CEO Sam Altman says society should accept some AI harm for the benefits and widespread availability of artificial intelligence. In an interview with POLITICO, Altman argued that some negative outcomes from AI should be tolerated rather than heavily restricting access to advanced systems. This view reflects an ongoing debate over how much risk is acceptable as AI becomes more powerful and prevalent.

Altman's stance highlights a divergence between OpenAI and rival Anthropic on how artificial intelligence should be regulated. While OpenAI believes the world should accept some "bad things" to allow people to exercise agency with AI, Anthropic has pushed for independent safety assessments. Altman drew a distinction between "bounded risks" and "catastrophic risks."

He said the latter warrants more caution but the former is acceptable in exchange for the advantages AI provides. The CEO noted that he anticipates AI will be used for "orders of magnitude more" good than harm. However, he did not consider every risk acceptable. Altman's comments indicate a philosophical divide between two leading AI developers, even as both advocate for greater scrutiny of cutting-edge models.

OpenAI recently agreed to support independent safety evaluations, while Anthropic has partnered with Accenture to include external evaluators. The disagreement is not simply about safety versus no safety. Rather, it centers on how much ordinary harm society should endure to maintain widespread access to powerful AI and where developers or policymakers should intervene.

As AI systems expand beyond text generation to act on users' behalf, businesses must weigh these risks. The abstract debate over acceptable AI risk becomes more concrete when AI can retrieve files, access data, write code, or interact with applications. Companies need to determine which risks are tolerable and which require stronger controls, especially when AI systems are authorized to take actions.

Altman's comments acknowledge that eliminating every possible AI-related harm may require accepting some trade-offs, but that does not mean businesses can simply outsource this decision to AI providers, regulators, or safety evaluators. The level of AI risk that is acceptable within an organization will depend on the system's capabilities.

For instance, an AI assistant summarizing internal meetings poses a different risk profile compared to an autonomous agent with permission to modify production code, access financial records, or interact with customers.

Written by urgent.news from TechRepublic's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

This story

This is one outlet's version. Read the fullest account.

Every outlet that reported this, and when →

Read the original at techrepublic.com →

More in AI

Recursive Governance: When Agents Write the Rules They Execute

We almost lost forty modules to a file that never changed. The sync job ran every six hours, mirroring our shared working directory into a test workspace. At 21:47 on a Tuesday, it removed about forty private modules that should have been ignored. The filter file was missing one pattern. That's it. One pattern.

  • Agents create and enforce their own rules, leading to conflicts and inconsistencies.
  • Recursive governance system protects modules and enables rule auditing and validation.

More from Tuesday 6 October →