Authenticator apps like Google Authenticator or Microsoft Authenticator are a stronger alternative.
In light of this threat, Google has published a new report on AI-powered cyber threats including how AI was used to hack.In the report, Google mentions that threat actors are leveraging AI to augment various phases of the attack lifecycle.
This exploit was built specifically to pick that lock.Wig says: "The attackers were not going after one person or one company.
AI now helps attackers write phishing messages that know your name, your employer, your role, and who your boss is.
These are not.Wig says that Google has already moved against the specific malware flagged in this report.
How was AI was used to hack the two-factor authentication process
Autonomous malware operations like PROMPTSPY were also used to hack
How could this exploit have impacted Indian consumers since it uses AI and is more advanced than before?
What can consumers do to stay safe from this exploit?
First, don't delay software updates. Only unpatched systems can be exploited using a zero-day exploit.
Second, move away from SMS OTPs where you can. This exploit bypassed software-based 2FA. Authenticator apps like Google Authenticator or Microsoft Authenticator are a stronger alternative. Some banks and apps now support them.
Third, go into your Android settings and check which apps have accessibility permissions. PROMPTSPY lives inside that access. If an app you barely use is listed there, remove it.
Fourth, treat unusually personalised messages with suspicion. If an email or WhatsApp message knows details about you that feel too specific, don't click. Call the organisation directly.
Finally, use different passwords across platforms. If one account is cracked, unique passwords ensure the damage stays contained. A password manager makes this practical without being painful.
Cyber hackers tried to hack two-factor authentication using a zero-day exploit in a Python script inside a 2FA system
Google Gemini AI tool was not used for this 2FA cyber attack
Suspected North Korea, Russia and China based hackers
Recently, Google 's Threat Intelligence Group detected the first known instance of a hacker group using an AI-developed zero-day exploit in the wild. This foreign hacker tried to hack the two-factor authentication process which many consumer applications, including those from banks and e-commerce, rely on for security.Google said that their own Gemini AI tool was not used for this hack, as the Python script used by the hackers was filled with educational docstrings, including a hallucinated CVSS score, and followed a structured, textbook Pythonic format highly characteristic of LLMs training data (like detailed help menus and the clean _C ANSI color class).While the attackers planned a wide-scale strike, Google's Threat Intelligence Group had deployed proactive counter-discovery, successfully stopping it from occurring. In light of this threat, Google has published a new report on AI-powered cyber threats including how AI was used to hack.In the report, Google mentions that threat actors are leveraging AI to augment various phases of the attack lifecycle. This includes supporting the development of vulnerability exploits and malware, facilitating autonomous execution of commands, enabling more targeted and well-researched reconnaissance, and improving the efficacy of social engineering and information operations.Google in the report said that Gemini AI tool was not used in this hack but rather some other AI tool was used.Tarun Wig, Co-founder & CEO, Innefu Labs, told ET Wealth Online that what Google caught this time is genuinely new territory.Wig says: "Their threat intelligence team found criminals who used AI to discover and build one of these exploits before anyone else spotted it. The plan was to run it at mass scale, hitting a huge number of systems at once. Google intervened before that could happen."Wig says that the target was the two-factor authentication process on a widely used server administration tool. That matters because 2FA is the safety net most people trust after their password.The AI found a flaw in how the developer had written the logic, a hidden contradiction buried in the code that traditional security scanners would never flag. Those tools look for crashes and errors. According to Wig, AI reads intent, and that is precisely what made this discovery so dangerous and different from anything that came before it.For those who aren't familiar, most software has bugs. Some bugs are harmless but some can be dangerous. But what's even more dangerous is zero-day.Wig says that zero-day is a type of bug no one is aware of yet, not the company that built the software, not the security teams watching it. Attackers who find one get a free pass into systems with no alarm going off.Google said in the report that AI-enabled malware, such as PROMPTSPY, signal a shift toward autonomous attack orchestration, where models interpret system states to dynamically generate commands and manipulate victim environments.In the report it was said that GTIG's analysis of this malware reveals previously unreported capabilities and use cases for its integration with AI. This approach allows threat actors to offload operational tasks to AI for scaled and adaptive activity.According to Wig, what researchers also uncovered is a separate Android malware called PROMPTSPY. It watches what you type on your own phone, learns your PIN or unlock pattern, and is built to resist deletion.Wig says: "Given that nine out of ten smartphones sold in India run Android, this is not a distant western cybersecurity problem. It is something sitting very close to home for the average Indian consumer."Think about how most Indians interact with money today. UPI payments, mobile banking, mutual fund apps, income tax portals, all of them sit behind an OTP. That OTP is the last lock on the door. This exploit was built specifically to pick that lock.Wig says: "The attackers were not going after one person or one company. The intent was mass exploitation, thousands or potentially millions of accounts in one sweep. For a country where digital payments crossed 18 billion transactions last year, the exposure would have been enormous."According to Wig, there is another layer that makes this particularly worrying for Indian users. AI now helps attackers write phishing messages that know your name, your employer, your role, and who your boss is. The old "Dear Customer" scams are easy to spot. These are not.Wig says that Google has already moved against the specific malware flagged in this report. Play Protect on Android is blocking known versions automatically.But the broader shift in how attacks are built means consumers need to take a few things seriously on their own. Wig tells what consumers can do:In the report, Google said that cyber crime threat actors remain interested in leveraging AI for vulnerability development as well.In one notable example, Google said that they noticed prominent cyber crime threat actors partnering to plan a mass vulnerability exploitation operation. Google's analysis of exploits associated with this campaign identified a zero-day vulnerability implemented in a Python script that enables the user to bypass two-factor authentication (2FA) on a popular open-source, web-based system administration tool.Google said in the report: "GTIG worked with the impacted vendor to responsibly disclose this vulnerability and disrupt this threat activity."Google says that they believe Gemini was not used, based on the structure and content of these exploits, and said that they have high confidence that the actor leveraged an AI model to support the discovery and weaponize the vulnerability.Google cited an example: The script contains a lot of educational docstrings, including a hallucinated CVSS score, and uses a structured, textbook Pythonic format highly characteristic of LLMs training data (e.g., detailed help menus and the clean _C ANSI color class).In the report, Google said that the vulnerability can be classified as a 2FA bypass, though it requires valid user credentials in the first place.Google said in the report that this hack stems not from common implementation errors like memory corruption or improper input sanitization, but a high-level semantic logic flaw where the developer hardcoded a trust assumption.According to the report, while fuzzers and static analysis tools are optimized to detect sinks and crashes, frontier LLMs excel at identifying these types of high-level flaws and hardcoded static anomalies.Google said: "Though frontier LLMs struggle to navigate complex enterprise authorization logic, they have an increasing ability to perform contextual reasoning, effectively reading the developer's intent to correlate the 2FA enforcement logic with the contradictions of its hardcoded exceptions."According to the report, this capability can allow models to surface dormant logic errors that appear functionally correct to traditional scanners but are strategically broken from a security perspective.Google said in the report that for the first time, GTIG (Google's Threat Intelligence Group) has identified a threat actor using a zero-day exploit that GTIG believes was developed with AI.The criminal threat actor planned to use it in a mass exploitation event but Google's proactive counter discovery may have prevented its use."Threat actors associated with the People's Republic of China (PRC) and the Democratic People's Republic of Korea (DPRK) have also demonstrated significant interest in capitalizing on AI for vulnerability discovery."The Google report pointed out that AI-driven coding has accelerated the development of infrastructure suites and polymorphic malware by adversaries.The report said: "These AI-enabled development cycles facilitate defense evasion by enabling the creation of obfuscation networks and the integration of AI-generated decoy logic in malware that we have linked to suspected Russia-nexus threat actors."
