Before the buzz, beyond the horizon
Competition pushes companies to release AI products sooner with no security in mind. Without designing fail-proof AI systems, companies put at risk their businesses, users, and society as a whole.
Increase trust in ML , AI and LLM applications to avoid cyberattacks or privacy issues and ensure safety.
Adversa AI experts are invited to comment attacks on AI, and our research results are published in top-tier media
“I would say most of the engineers working on A.I., they don’t understand the new attack vectors,” Alex Polyakov, the founder and CEO of Israeli A.I. security startup Adversa.Al., says.
What can we do to minimize the harm from AI? We must understand that we’re creating a new creature that will have great power beyond our own. …if we don’t teach and train it correctly from the very beginning, it can make things worse than they are now.
“Research from cybersecurity and safety firm Adversa AI indicates GPTs will leak data about how they were built, including the source documents used to teach them, merely by asking the GPT some questions.”
Adversa AI’s technique is designed to fool facial recognition algorithms into identifying a picture of one person’s face as that of someone else by adding minute alterations…
“Content-based verification might be one of the first victims of real cyberattacks, because it grants direct access to critical information,” says Alex Polyakov, the founder and CEO of Adversa AI, a company dedicated to researching trustworthy AI.
“According to the new discovery by researchers at Adversa AI, the vulnerabilities affect custom GPT applications built on OpenAI’s powerful API.”
…Adversa AI published a blog post demonstrating methods to bypass OpenAI’s content filters and get GPT-4 to generate phishing emails, offensive descriptions of gay people and other highly objectionable text.
“Hackers are pioneering new ways of tricking facial-recognition systems, from cutting the eyes out of photos to making a portrait ‘nod’ with artificial intelligence.”
“The research highlights the vulnerability of existing AI safety filters and should serve as a wake-up call for the AI community to bolster security measures across the board, says Alex Polyakov, co-founder and CEO of security company Adversa AI.”
…such ‘toy’ jailbreak examples will be used to perform actual criminal activities and cyberattacks, which will be extremely hard to detect and prevent,” Polyakov and Adversa AI write in a blog post detailing the research.
These sorts of attacks [adversarial AI attacks] are not yet numerous, but have happened, and will happen with greater frequency, says Eugene Neelou, co-founder and CTO of Adversa AI.
“Adversa aim to automate red teaming activities to help organizations investigate the robustness of their guardrails.”
Adversa AI is collaborating with technology and industry partners around the world to advance the mission of Secure and Safe AI
Our mission is to help vendors build and monitor secure and trustworthy AI systems
It’s time to protect your AI systems. Get in touch with us!
