\u201cAI shows promise in enhancing cybersecurity by modeling complex threats and identifying vulnerabilities, but real-world effectiveness depends on precise problem definitions, quality data, human oversight and addressing security risks like adversarial attacks, membership inference attacks and poisoning attacks,\u201d Zhang said.<\/p>\n
Enhancing cybersecurity through AI<\/strong><\/h3>\nResearchers have had success adopting mathematical formulas of cybersecurity challenging and then using AI to solve those formulas. Zhang said that in lateral movement attacks, which enable attackers to move through a compromised network after gaining initial access, researchers can model a network as a graph. AI has shown strong potential for detection and defense in these instances.<\/p>\n
That\u2019s research, though, not a real attack. Large language models (LLMs) can successfully identify and fix bugs in programs with fewer than 100 lines of code, but as the database gets larger, the LLM has to be taught with much greater specificity. Human oversight is essential.<\/p>\n
\u201cReal-world environments are far more complex than controlled scenarios,\u201d Zhang said. \u201cEffective deployment of AI in practice requires more precise problem formulations and algorithms tailored to specific challenges. AI is not a magic solution\u2014it depends on the availability of high-quality training data, well-scoped problem definitions and effective learning methodologies.\u201d<\/p>\n
AI in your daily life<\/strong><\/h3>\nLong before ChatGPT, most of us interacted with AI somewhat regularly\u2014talking to Siri and Alexa, using our faces to unlock our smartphones, autopilot in cars. They\u2019re valuable but also pose security risks for users\u2014facial recognition can be fooled, Tesla\u2019s AI can misread altered traffic signals and Siri can be trigged by hidden commands in audio.<\/p>\n
\u201cAs reliance on AI grows, so do the risks, making security and robustness critical for protecting everyday users,\u201d Zhang said.<\/p>\n
Her research is looking into security gaps coming up as people increasingly rely on LLMs for decision-making and tech support.<\/p>\n
Security risks in AI systems<\/strong><\/h3>\nThere are established algorithmic biases in AI, such as racism or sexism in responses, but security challenges extend beyond that.<\/p>\n
\u201cBlind spots can be exploited by adversaries and pose real risks to the general public,\u201d Zhang said. \u201cOur current research aims to identify and understand these AI blind spots, with the long-term goal of building more secure, resilient and robust models that can withstand adversarial manipulations.\u201d<\/p>\n
There\u2019s also a risk of jailbreaking LLMs, which refers to manipulating the system into bypassing their built-in safety constraints. Zhang said a model like ChatGPT can be \u201ctricked\u201d\u2014it\u2019s prohibited from offering instructions on building explosives, but attackers can embed malicious prompts within innocuous-looking text to trick the LLM into generating prohibited content. Hackers also use this method in conducting cyberattacks.<\/p>\n
![\"Northern](\"http:\/\/in.nau.edu\/news\/wordpresst\/uploads\/sites\/153\/wp-content\/uploads\/2019\/06\/NAU_primary-281_3514-300x213.png\")
<\/p>\n
Heidi Toth | NAU Communications
\n(928) 523-8737 | heidi.toth@nau.edu<\/a><\/p>\n","protected":false},"excerpt":{"rendered":"
Long before ChatGPT, most of us interacted with AI somewhat regularly\u2014talking to Siri and Alexa, using our faces to unlock our smartphones, autopilot in cars. They\u2019re valuable but also pose security risks for users\u2014facial recognition can be fooled, Tesla\u2019s AI can misread altered traffic signals and Siri can be trigged by hidden commands in audio.<\/p>\n
\u201cAs reliance on AI grows, so do the risks, making security and robustness critical for protecting everyday users,\u201d Zhang said.<\/p>\n
Her research is looking into security gaps coming up as people increasingly rely on LLMs for decision-making and tech support.<\/p>\n
Security risks in AI systems<\/strong><\/h3>\nThere are established algorithmic biases in AI, such as racism or sexism in responses, but security challenges extend beyond that.<\/p>\n
\u201cBlind spots can be exploited by adversaries and pose real risks to the general public,\u201d Zhang said. \u201cOur current research aims to identify and understand these AI blind spots, with the long-term goal of building more secure, resilient and robust models that can withstand adversarial manipulations.\u201d<\/p>\n
There\u2019s also a risk of jailbreaking LLMs, which refers to manipulating the system into bypassing their built-in safety constraints. Zhang said a model like ChatGPT can be \u201ctricked\u201d\u2014it\u2019s prohibited from offering instructions on building explosives, but attackers can embed malicious prompts within innocuous-looking text to trick the LLM into generating prohibited content. Hackers also use this method in conducting cyberattacks.<\/p>\n
<\/p>\n
Heidi Toth | NAU Communications
\n(928) 523-8737 | heidi.toth@nau.edu<\/a><\/p>\n","protected":false},"excerpt":{"rendered":"
<\/p>\n\n(928) 523-8737 | heidi.toth@nau.edu<\/a><\/p>\n","protected":false},"excerpt":{"rendered":"