▲ OpenAI
OpenAI's new artificial intelligence (AI) model, "GPT-6 Astra," has successfully passed all stages of the "Captcha" games used to prove human identity on the internet.
Sharif Shameem, a developer at OpenAI, announced on social media X on the 8th (local time) that after testing GPT-6 Astra on the Captcha game "I'm Not a Robot," the AI solved all 48 stages of the game and obtained a "human certificate."
"I'm Not a Robot" consists of games such as guessing distorted alphabets, distinguishing between cookies and dogs in similar-looking photos, and identifying the locations of specific structures like traffic lights within images.
These are games currently in actual use to filter out crawling bots that scrape websites on the internet, and some are known to have high difficulty levels that often cause even real humans to give incorrect answers.
Astra's ability to solve all the Captcha games is attributed to its advancement to a level similar to humans across various domains, including spatio-temporal reasoning capabilities like image recognition, contextual reasoning capabilities such as instruction comprehension, and computer usage skills.
As commercial AI models neutralize tools that have been used to distinguish between humans and robots in this manner, a major overhaul of robot-blocking technology is expected to become necessary.
Meanwhile, Artificial Analysis, an AI model performance analysis firm, revised the evaluation system of its Artificial Analysis Intelligence Index (AAII) twice over four days, raising Astra's ranking.
On the 4th, the day of its release, Astra scored 61 points under the previous evaluation system (version 4.1), entering the rankings tied for 5th place behind Anthropic's "Claude Fable 5.1" (66 points) and others. However, after applying the new system (version 4.3), it tied for 1st place with Fable 5.1 with a score of 53 points.
Artificial Analysis explained that it preemptively applied some features of the upcoming "version 5" in line with the advancement patterns of cutting-edge AI technology.