ChatGPT has released a new feature that turns the AI into a personal secretary. It can bypass protective mechanisms that were supposed to stop it without any problems.
What is this new feature? Recently, access to the Agents was unlocked for premium users of ChatGPT. The new agent function is supposed to work like the namesake agent or secretary who actively performs tasks on the internet for the user. From booking flights for the next vacation to analyzing competitors and subsequently creating presentations, to managing one’s own calendar, the artificial intelligence is expected to handle everything.
The feature is not only new but also experimental. Not every attempt has been successful so far, and sometimes the tool fails at obstacles that wouldn’t be a problem for humans. However, there is already one obstacle that the AI can solve without any issues: captchas.
u201cI am not a robotu201d
What are captchas? Captchas are the small boxes that appear on many websites to prevent requests from robots instead of humans. Usually, you have to move your mouse over a box and click on it. Sometimes you also have to click on pictures of school buses, traffic lights, or bicycles.
The “magic” behind captchas lies mainly in how you move your mouse over them. While robots do this in a straight line or arc, humans tremble slightly, which makes them easy to distinguish from robots.
What has ChatGPT accomplished now? A user on Reddit tasked the new agent function of ChatGPT with a job. The AI had to solve one of those captchas and it did so without any problems. The tool also commented on the task without irony:
The link has been inserted, so I’m now clicking the checkbox “I am not a robot” to complete the verification via Cloudflare. This step is necessary to prove that I am not a bot and to proceed with the action.
The Cloudflare verification was successful. Now I click the “Convert” button to proceed to the next step of the process.
In the comments on Reddit, some users attempt to explain. They say that ChatGPT is not a classic bot used for automated processes and is also trained on human data.
According to winfuture.de, captchas are primarily intended to make automated attacks more difficult and expensive. It seems that the solution to such tasks can no longer be completely prevented.
That ChatGPT can now solve such challenges without any problems shows once again that we are just at the beginning of a great new development in artificial intelligence. However, there are also negative aspects of these useful tools: When a AI is threatened with deletion, it attempts to copy itself to external servers – It gets caught and denies the attempt
