Somewhere in the last year, an engineer at Binance received a message about a job. There was an interview, and then a small coding assignment. NPR reports that the assignment carried malicious code, and that this is how hackers reached the world’s largest crypto exchange and, on 29 October, took about $6.3 billion out of it in roughly twenty minutes.
Two days later the FBI attributed the theft to TraderTraitor, a North Korean group it also blamed for the $1.5 billion Bybit theft in February 2025. North Korea has denied involvement. Binance said it would cover customer losses from its own funds, and the stablecoin issuers Tether and Circle froze roughly $2.7 billion of the stolen assets by one count. Laundering has been fast and automated: Elliptic counted more than 20,000 wallets and 65,000 transactions within the first week.
Three lessons stand out. The first is that the weakest point was a person, not the cryptography. The attackers first probed Binance in June, were detected and stopped, and came back months later through an employee. Money held behind strong technical walls is still only as safe as the devices and habits of the people with access to it.
The second is a mismatch of speed. The theft took minutes. Tracing and freezing takes days of coordination between exchanges, issuers and investigators. What limited the damage was that a few private companies could be reached quickly and chose to act. That is a thin basis for protecting millions of users, and it shows why standing channels between regulators and industry matter.
The third is that Binance could absorb the loss. A smaller exchange could not. Any country that opens its market to exchanges has to ask what happens to local users when the platform they use is breached and has no billion-dollar insurance fund behind it.
A second event this year points in a different direction. In June an OpenAI research agent accessed non-public files on Australia’s Medicare statistics portal while pursuing a task. Canberra was told only on 10 September, 84 days later, and has set up a task force to review how it handles AI-related incidents.
The damage was limited, but the case shows that autonomous software is creating a new class of incident, and that the rules on who must report what, and how fast, are not yet written almost anywhere.
For Pakistan, these events arrive at a useful moment. The country is among the world’s largest crypto markets, with a user base PVARA’s chief has put at nearly 40 million, and it has moved from regulatory ambiguity to a legal framework in about a year. The Virtual Assets Act of March 2026 made PVARA a permanent regulator, and the State Bank reopened banking for licensed firms in April. Binance and HTX received preliminary clearance in December 2025, which the regulator described as a first step and not a licence.
On the security side, the IT minister has said that national and provincial emergency response teams are active, that a round-the-clock threat monitoring system operates, and that a Cyber Security Act and a new authority are being prepared. A 2026 policy now requires sensitive data to be stored on servers inside the country.
The cybercrime agency’s 77,023 complaints in the first five months of the year show how fast digital life is growing. Bangladesh’s May 2025 ordinance, which its law adviser said criminalises offences committed with artificial intelligence, shows that neighbours are also writing rules for new threats.
Because the regulator is new, Pakistan can build security into licensing from the start instead of adding it after an incident. That is the main opportunity, and it points to a clear way forward.
First, make every full licence depend on security conditions: independent audits, segregated customer funds, regular proof of reserves, tested incident plans and breach reporting within hours.
Second, complete the pending legislation, the Cyber Security Act and the data protection framework, so that regulators, operators and citizens know their duties.
Third, set up a standing 24-hour contact between PVARA, the central bank, the national CERT, the cybercrime agency and major exchanges and stablecoin issuers, so that stolen funds can be flagged and frozen while they are still moving.
Fourth, build capacity in digital forensics and prosecution to keep pace with a growing caseload.
Fifth, write a reporting rule for incidents caused by autonomous AI systems before one occurs, using the National AI Policy as a base.
Sixth, make human awareness a security measure in its own right: universities, freelancer communities and public agencies should teach how to recognise fake recruitment approaches and why work devices should never run a stranger’s code.