Google to expand AI-powered fraud detection and security operations in India

Google has announced its safety charter in India. This will expand the company’s largest market outside the US, the AIRE LED development to combat fraud across the country.

Digital fraud in India is on the rise. Fraud related to the Indian government’s immediate payment system An increase of 85% from the previous year Last year, the Indian rupee reached nearly 11 billion ($127 million), according to government data. India has also seen several instances of digital arrest fraud. There, scammers were raised as officials to force money through video calls and predatory loan apps.

Google aims to address some of these areas with its safety charter. The company launched it Security Engineering Center It is the fourth centre in India after Dublin, Munich and Malaga.

Announced last year at Google for India Summit, the Security Engineering Center (GSEC) allows Google to partner with local communities, including governments, academia, students and small businesses, and create solutions to solve cybersecurity, privacy, safety and AI issues.

Google is partnering with the Ministry of Home Affairs’ Cybercrime Coordination Centre (I4C) to raise awareness of cybercrime, the company said in a blog post. This is based on the company’s existing work, including the launch of an online fraud identification program. Digikavachdebuted in 2023, limiting the harmful effects of malicious financial and predatory loan apps.

With GSEC in India, Google will focus on three key areas, Adkins told TechCrunch. The phenomenon of online fraud and fraud and how people are safe online. Cybersecurity for businesses, governments, and critical infrastructure. Building responsible AI.

“These three areas will become part of India’s safety charter and we hope to use the fact that in the coming years we have engineering capabilities here to solve what’s going on in India, where users are,” Adkins said.

Globally, Google is using AI to fight online fraud and remove millions of ads and ad accounts. The company aims to deploy AI more broadly in India to combat digital fraud.

Google Messages, pre-installed on many Android devices, uses AI-powered fraud detection that helps protect users from over 500 million suspicious messages a month. Similarly, Google I piloted the protection of the play India last year claims it was blocking nearly 60 million attempts to install high-risk apps, with over 220,000 unique apps halted on more than 13 million devices. Google Pay, that’s true One of the top UPI-based payment apps There was also a 41 million warnings in the country for transactions suspected of being a potential fraud.

–

Adkins, a founding member of Google’s security team, who has been part of an internet company for over 23 years, discussed several other topics in an interview with TechCrunch.

Adkins said one of the best things about the mind is the use and misuse of AI by malicious actors.

“We obviously track AI very closely, and up until now we’ve used large-scale language models like Gemini as productivity gains. For example, to make phishing scams a little more effective, we’ll use the advantages of translation to make them more reliable, especially if the actor and target have different languages.”

Adkins said it is doing extensive testing of its AI models to ensure that it understands what it should not do.

“This is important for generated content that may be harmful, but it is also important for the actions it may take,” Akins says.

Google is working on a framework that includes Ensure an AI frameworklimits the abuse of its Gemini model. However, to protect generative AI from being abused and abused by hackers in the future, the company believes it needs a framework to build safety for the communication of multiple agents.

“The industry is outputting protocols very quickly (and thus), which seems like the early days of the internet, where everyone releases code in real time, thinking about safety after the fact,” Adkins said.

Google simply doesn’t want to introduce its own framework to limit the scope of generated AI being abused by hackers. Instead, Adkins said the company is working with the research community and developers.

“One thing you don’t want to do is to constrain yourself too much in your early research days,” Adkins said.

About monitoring vendors

In addition to the potential for hacker-generated AI abuse, Adkins sees commercial surveillance vendors as a serious threat. These can include spyware manufacturers including NSO Group,That one Notorious for Pegasus Spywareor other small businesses selling monitoring tools.

“These are companies spun around the world, developing, manufacturing and selling platforms for hacking,” Adkins said. “Depending on the refinement of the platform, you might pay $20. You could pay $200,000. You could also expand those who attack without their own expertise.”

Some of these vendors also sell tools Spy on people in the market including India. However, apart from being targeted by surveillance tools, the country faces its own challenges in terms of its size. The country is looking at AIRED’s deepfakes and voice cloning scams as well as instances of Digital arrestAdkins’ emphasis is a regular scam adapted to the digital world.

“You can see how fast the threat actors themselves are moving forward, which is why I love studying cyber in this region. This is a hint of what you can see around the world at some point,” Adkins said.

About multifactor authentication

For a long time, Google has encouraged users to protect their online presence by using more secure authentication methods beyond their passwords. The company has also turned on Multifactor Authentication (MFA) for all past user accounts, and has also promoted hardware-based security keys. The same goes for passwordless It becomes a popular technical termhas many different meanings.

Nevertheless, it is difficult to expect people to abandon their passwords in a market like India due to their vast demographics and diverse economic landscapes.

“For a very long time, we knew that passwords were insecure. This concept of multifactor authentication took a step forward,” Adkins said, adding that Indians are more likely to support SMS-based authentication than other MFA options.