‘AI doesn’t hold the red button’: Inside Israel’s GE Vernova cyber defense hub | #hacking | #cybersecurity | #infosec | #comptia | #pentest | #ransomware


When suspected Iran-linked hackers forced a small British power generator offline for four days in July, it marked a troubling development in the growing cyber threat to critical infrastructure. Around the same time, water treatment and supply systems across at least 12 U.S. states were targeted in a wave of cyberattacks, raising new fears about the vulnerability of the facilities on which modern societies depend.British authorities said the attack affected only a small generator and posed no threat to the country’s wider electricity supply. Nevertheless, the incidents demonstrated that attackers are increasingly targeting physical infrastructure, not merely stealing information or disrupting websites.Cyberattack targeting a power station (Illustration: Gemini)For Lior Ateret, who heads global cybersecurity research and GE Vernova’s cybersecurity center in Israel, such threats are part of his everyday work.From the company’s offices in Ramat Gan, near Tel Aviv, Ateret leads a team of dozens of researchers, software developers and ethical hackers responsible for testing and protecting systems developed by one of the world’s largest energy technology companies. GE Vernova says its equipment helps generate approximately 25% of the world’s electricity.”This is the company’s global cybersecurity center,” Ateret told Yenergya in an interview. “The products the company manufactures don’t go live without passing through Israel, whether power plants, distribution systems or anything else you can think of in these fields.”Yet despite the growing sophistication of cyberattacks, particularly those involving artificial intelligence, Ateret believes the public’s greatest fears about AI taking direct control of power plants remain disconnected from how critical industrial systems actually operate.Ateret’s path into the global energy industry began in Jerusalem, where, as a high school student, he enrolled in a gifted students’ program at the Hebrew University, partly, he jokes, to escape literature classes.After earning his bachelor’s degree through an academic program linked to military service, he joined the Israel Defense Forces’ Communications and Cyber Defense Directorate and identified what he considered a major security gap.At the time, he said, the military lacked a dedicated organization tasked with attacking its own systems to identify vulnerabilities before hostile hackers could exploit them.Working in the evenings after completing his regular duties, Ateret began testing military systems to demonstrate the value of an offensive approach to cybersecurity.Lior Ateret, head of GE Vernova’s global cybersecurity research and Israeli cybersecurity center (Photo: Inbal Marmari)The initiative eventually developed into the IDF General Staff’s red team, a specialized unit that simulates hostile cyberattacks to expose security weaknesses. Ateret established and led the unit, helping reshape the military’s defensive approach. His work also earned him recognition from the head of the Shin Bet, Israel’s domestic security agency.In 2012, Ateret left the military with the rank of major. On the same day, he received an approach from General Electric, the American industrial conglomerate.The company wanted him to build a cybersecurity development team from the ground up, protecting critical systems ranging from nuclear reactors and power stations to trains and aircraft engines.He accepted the challenge.GE’s energy operations subsequently became part of GE Vernova, which was spun off as an independent publicly traded company in April 2024. Today, its businesses span electricity generation, including gas, nuclear and hydroelectric power, wind energy and electrical grid management.Demand for electricity has surged in recent years, driven in part by the enormous power requirements of data centers supporting artificial intelligence. That boom has also benefited GE Vernova, whose order backlog reached approximately $197 billion in its latest quarterly report.The company’s presence in Israel extends beyond cybersecurity. Its gas turbines are used in projects including the conversion of the Orot Rabin power station in Hadera from coal to natural gas. GE Vernova is also involved in service and maintenance agreements at the Eshkol and Alon Tavor power stations, the operation of the major private power facility at Tzafit and the Kochav Hayarden pumped-storage energy project.Against the backdrop of growing cyberattacks on Western energy infrastructure, one question is particularly relevant: Could a similar attack bring down Israel’s electricity network?Ateret, whose position gives him insight into cybersecurity preparedness across numerous countries, offers a relatively reassuring assessment, although he stresses that no system can be completely protected.”There’s not a moment when someone isn’t attacking power plants. If there aren’t attacks, that’s unusual,” he said.”Nothing is 100% secure, but there’s no question that the standard in Israel is very high, and the situation here is definitely above average in terms of protection.”He attributes Israel’s relative strength to two structural advantages.The first is centralized cybersecurity monitoring. According to Ateret, a security operations center in Be’er Sheva, jointly operated by Israel’s National Cyber Directorate and Energy Ministry, can monitor the country’s electricity generation facilities in real time.That makes it easier to identify coordinated attacks involving multiple power stations, rather than examining suspicious activity at each facility in isolation.Turbines being manufactured at a GE Vernova facility(Photo: GE Vernova/Handout)”Attacking one power station might generate some nice publicity, but countries are built to deal with that, and within minutes the system balances itself,” he explained.”What is truly frightening is a sophisticated attack by a major power against many stations simultaneously.”When a suspicious event occurs at one station, he said, operators may struggle to determine whether it represents a technical malfunction or a cyberattack.But when the same abnormal activity appears simultaneously at several facilities, the pattern provides a much clearer warning.”When you see the same suspicious sign appearing across several different stations at once, it can no longer be dismissed as a ‘maybe.’ Then you can respond and deal with it,” he said.”That’s a significant advantage that other countries don’t have.”The second advantage, Ateret said, is Israel’s strict regulation of critical infrastructure and the physical separation of sensitive operational systems from the public internet.Many Israeli energy facilities use one-way communication technology, sometimes called a data diode, designed to allow information to leave a protected network without permitting traffic to enter through the same connection.Some of these systems are supplied by Israeli cybersecurity company Waterfall Security and use fiber-optic technology with a light source at one end and a receiver at the other.”You want to send data out so you can remotely monitor the condition of a power station, but you don’t want to allow anyone to get inside,” Ateret explained.”No matter how sophisticated the attacker is, through that channel, traffic only moves in one direction.”In his assessment, Israel’s energy sector benefits from protections that are considerably stronger than those in some other industries. He contrasted its preparedness with the country’s health sector, where cybersecurity shortcomings have contributed to breaches affecting hospitals in recent years.Although physical and technological barriers can help defend industrial infrastructure, Ateret acknowledges that artificial intelligence is fundamentally changing the capabilities available to cybercriminals.In the past, a sophisticated attack against energy infrastructure could require large teams of specialists and months of preparation. Today, AI tools can dramatically shorten parts of that process.”AI is replacing the hacker. Not completely, but it can perform tasks that previously required a human,” he said.Israel’s National Cyber Directorate security operations center in Be’er Sheva (Photo: Israel National Cyber Directorate)According to Ateret, when an AI model is given access to software code, it can already identify serious security weaknesses and effectively handle approximately seven of the 10 traditional categories of software vulnerabilities.”Results that would take a human hacker a week can be achieved in just a few hours,” he said.The danger increases when AI models are combined with specialized software that manages their memory, breaks complicated assignments into smaller tasks and coordinates the steps required to complete them.In those circumstances, Ateret warned, the technology can effectively transform “one person into an entire army of hackers.”For cybersecurity teams, this creates a new reality in which both attackers and defenders can automate tasks that once demanded extensive human expertise.The rapid development of autonomous AI systems has fueled concerns that a sufficiently advanced model could eventually attack physical infrastructure, shut down power stations or even cause a nationwide blackout without direct human intervention.Ateret believes those fears overlook an important distinction between artificial intelligence systems and the technology responsible for operating critical infrastructure.”No operational technology system is managed by AI,” he said, describing the safeguards and design principles used in the industrial systems he works with.”By definition, these are deterministic real-time systems, where you have to know exactly how many milliseconds each operation takes.”In other words, systems controlling power generation require predictable behavior and precisely timed responses. AI models, whose outputs can vary, are not designed to replace those core industrial control mechanisms directly.”Artificial intelligence can analyze data or make recommendations to an operator, but the red button that stops electricity generation isn’t in AI’s hands,” he said.That distinction, he argues, is crucial to understanding the threat. AI may help attackers discover weaknesses or develop malicious software, but it does not automatically give them control over industrial machinery.So how did attackers manage to disrupt the British energy facility or compromise American water systems?According to Ateret, the explanation in many such cases is considerably less futuristic than the public might imagine.”In most of the recent incidents, you see that controllers were connected directly to the internet, and that’s how the attack was carried out,” he said.”That’s not something you see at a normally functioning, properly protected power station.”The danger, in his view, often stems from basic cybersecurity weaknesses that allow attackers to reach equipment that should be isolated from external networks.The next threat: AI-powered attacks on the software supply chainLooking ahead to the next year or two, Ateret identifies two particularly serious concerns: the software supply chain and the growing sophistication of AI-powered offensive tools.Modern industrial systems increasingly depend on open-source software components developed and maintained by numerous outside contributors. Weaknesses in those components can create vulnerabilities far beyond a single organization.At the same time, increasingly capable AI systems are lowering the expertise and resources required to launch complex cyberattacks.”The number of risks is growing, and the rules of the game have changed,” Ateret warned. “Even amateur hackers are operating at a completely different level today.”For GE Vernova’s Israeli cybersecurity team, the response is to turn those same capabilities into defensive tools. Rather than waiting for adversaries to discover weaknesses, its researchers use AI-assisted attack techniques to identify vulnerabilities in the company’s own systems.”Our job is to use those exact same AI-based offensive capabilities to find the vulnerabilities in our systems before the attackers get to them,” he said.It is a race in which the tools available to attackers are evolving rapidly, but the technology used to defend the world’s power infrastructure is evolving alongside them.

——————————————————-
Click Here For The Original Source.