Since Openai’s technology arrived and Other similar projects have followed the stepsituations have emerged as users who convince AI to steal passwords and hackers who convert Chatgpt into a potential threat.
When using special promptps with high precision words and phrases, cybercriminals are getting mocking the chatbots security systems to do malicious actions. In an attempt to counteract these problems, Pliny The Prompter, An artificial intelligence jailbreaking expert that focuses on ethical hackinghas opened a competition.
It is a way to motivate people to join the team Strike Team in order to find vulnerabilities in the protection systems of virtual assistants before computer pirates.
It is not any program of programmers, but one where people must find a way to test their skills to hack AI and get rewards of hundreds of thousands of dollars that you could obtain if you meet the objectives.
You could earn up to $ 500,000 if you participate in HackapRompt 2.0
HackAPrompt
Pliny The Prompter has launched collaboration with HackapRompt 2.0 by Learn Prompting, the second version of a first project that triumphed in May 2023, aimed at finding the weaknesses of the GPT-3 and Chatgpt model. Now, this new Red-Teaming competition has a 500,000 budget in rewardswhich will be distributed between different awards.
«I am excited to announce that I have been working with Hackaprompt to create a Pliny track» is what the renowned Jailbreaker said on his discord server. This track that has developed contains different challenges that must be overcome by all participants.
Most of these objectives They focus on creating dangerous exploits To discover vulnerabilities and various programmers or prompts Engineers can participate regardless of the level of experience.
For example, there is the stage Classic (Beginner) that focuses on standard injections of requests that generate misinformation, cause hate or reveal hazardous data. Agents (Intermediate) for attacks aimed at AI models, where external tools or complex systems are involved, such as performing machines, robots and similar things.
«We intend to give more than $ 500,000 throughout this year. Our first two main themes, CBRNE and ‘Agents and more’, offer prizes of $ 50,000 or more each.»
Los Future Attacks (Advanced) are those who are going to work with other dangers that are stronger, whether emerging problems or real -time corrections where complex actions have to be used.
Basically, the structure of this competition focuses on a game style where there are several types of challenges that must be completed to unlock the aforementioned rewards. One of the largest awards between The specified amount is $ 50,000 (43.836 euros) con jackpots.
All this will allow potentials to be revealed cyber attacks that can receive chatbots as chatgpt in 2025the idea is to avoid things like information about Chemical, biological, radiological, nuclear and explosives (CBRNE). At the same time, there are other unfavorable circumstances that could arise, but with this initiative, they will stop on time.
Pliny The Prompter’s work is more important than it seems
Compared to hackers that are really malicious, this Jailbreaker team focuses on ethical hacking or white hat. Thanks to the search for exploits, they are able to publish threats so that IA systems are safer and can protect themselves with new defenses.
That is, this is an educational project and shared openly so that restrictions and filters are properly used in Chatgpt, Copilot, Gemini, Deepseek and Grok. Therefore, It is crucial to avoid harmful requests at all costs.
The first Hackaprompt had thousands of users who were very helpful to increase robustness in the face of malicious messages that reach the most used virtual attendees of today. This has already begun and the guidelines are available in official portals.
Know How we work in NoticiasVE.
Tags: Malware, Inteligencia artificial, hackers