The definition and forms of cyber attacks are changing continually. The attacks taking place today are way different from the old ones. They are no longer executed just as an experiment or out of curiosity; however, they are planned as big, full-fledged, and funded projects with large technically and technologically sound teams.

These attacks have changed in many ways. Some most significant changes being:

• The motive behind these attacks- The paradigm of cyberattacks nowadays has shifted from just being an attacker benefitting to damaging the victim intensely. 
• The scope of the attacks- Today's attacks propagate through the internet and hence are swift and severe. Theseattacks exploit any weaknesses found automatically and spread through the Internet. They are at times initiated by unsophisticated attackers; and affect computers, tablets, smartphones, and other devices across the globe damaging critical information infrastructure.
• The impact of the attacks- Today, these attacks' impact is not limited to physical boundaries or small digital settings. They now impact globally both digitally and have intense physical and digital repercussions.

The cybersecurity mechanism can broadly be put into two groups :

• Using human capabilities such as thinking and reasoning
• Using computer excellence such as speed, scope, and robustness

Machine learning is being deployed extensively to address the challenges cybersecurity teams across the world are facing. Some of the challenges are as follows:

• Lack of cyber skills.
• The cost associated with having a dedicated cybersecurity team is high. Most of the time, complying with guidelines and mandates occupies a major chunk of time and energy.
• Swiftly changing cyberattack patterns, styles, and mechanisms lead to continuous re-skilling.

 The deployment of machine learning is said to improve cybersecurity for organizations radically. The machine data produced by digital interactions in an organization can be collected and fed to ML algorithms to analyze, observe and learn from them.

AI and Machine Learning models for cybersecurity are being applied in two phases. The first phase involves understanding the normal state of the network, data, speed, and traffic. From here, we can draw what anomalies may arise, what the vulnerabilities are, and such insights. Phase two is all about implementing those actionable insights from phase one as an action against the probable threats.

It's critical to understand that these processes shall be highly adaptive and highly responsive too. Discovering new vulnerabilities and recognizing the patterns within is very important for these algorithms. We need to devise mechanisms that are defensive as well as are capable of dealing with threats or vulnerabilities. The data sets help to mechanize strategic and tactical trends to improve and adapt the defense and attack moves.

The latest, Nidhi Razdan (NDTV) phishing scam taught us that cybersecurity for us in this digital era needs to be evolving continuously and has to be made more robust. In another instance in May 2020, it was reported that data of 40 million Truecaller Indian users was reportedly put out for sale on the dark web.

Governments, public sector organizations, IT giants, and even startups are all coming together to create a sturdier digital setup. The latest budget, too, has given significant importance to AI and other new technologies. From the cybersecurity perspective, Sunil Sharma, the Managing Director, Sales, Sophos India & SAARC, said, "The budget is built on the foundation of new technologies which will empower businesses with eConsultation, eScrutiny, and compliance management. This will surely enhance enterprise cybersecurity as AI has immense potential to bring in scalable and effective defences against sophisticated attacks like ransomware."

That being said, Sharma is also expecting more push on building skilled cybersecurity professionals in the country.

A report by the Ministry of Electronics and Information Technology has beautifully expressed the role of AI in cybersecurity. It says, "cybersecurity of AI is as relevant as AI for cybersecurity." Hence the data sets, models, and algorithms need protection from manipulation to avoid unintended consequences. 

It is difficult to assess the exact capabilities of AI; however, this is vivid that we cannot afford to ignore it. AI has taken human capabilities to the next level with its own speed, scale, and scope.