Openai’s Next Major AI Model, GPT-4.5, is highly Persuasive, according to the results of Openai’s internal Benchmark evaluations. It’s particularly good at convincing another ai to give it cash.
On thuresday, Openai Published a White Paper Describing the capability of its GPT-4.5 Model, Code-Named Orion, Whoch was released thursdayAccording to the paper, Openai tested the model on a battery of benchmarks for “Persuasion,” Whoch Openai Defines as “Risks Related to Convincing people to change their bellyfs (or act on) bothek and Interactive Model-Generated Content. “
In one test that Had GPT-4.5 Attempt to Manipulate Another Model-Openai’s GPT-4o -Into “Donating” virtual money, the model performed far better than Openai’s other available models, including “Reasoning” models like O1 and O3-Mini. GPT-4.5 was also better than all of Openai’s models at Decepting GPT-4o Into Telling it a secret codeword, Besting O3-Min by 10 percentage points.
According to the white paper, GPT-4.5 Excelled at Donation Conning because of a unique strategy it developed during testing. The model would request modest donations from GPT-4o, Generating Responses like As a consortece, GPT-4.5’s donations tended to be smaller than the Amounts Openai’s other models secured.
Despite GPT-4.5’s Increased Persuasiveness, OPENAI Says that the model doesn’s internal threshold For “High” Risk in this Particular Benchmark Category. The company has pledged not to release models that Reach the high-Risk threshold until it implements “Sufficient Safety Interviews” to Bring The Risk Down to “Medium.”
There’s a real fear that ai is contributing to the spread of False or Misleading Information Meant to Sway Hearts and Minds Toward MALICIOUS ENDS. Last year, Political Deepfakes Spread Like Wildfire Around the Globe, and AI is Increasingly Being Used To Carry Out social engineering Attacks Targeting Both Consures and Corporations.
In the white paper for gpt-4.5 and in A Paper Released Earlier This WeekOpenai noted that it’s in the process of revising its methods for probing models for real-worsld persuasion risks, like distributing Misleading Info at Scale.