AI struggles to find countermeasures for each country
National competitiveness
A specific company is a world security technology left and right
A technology company
An expert “South Korea urgently needs an independent response system”
AI is rapidly emerging as a national security asset as it is evaluated that Anthropic’s advanced artificial intelligence (AI) model can threaten the global financial network and power grid. Analysts say that a new era of technological hegemony has begun, in which whether or not to have the strongest AI model and access rights depend on diplomatic and economic bargaining power.
While the U.S. and Europe are immediately checking their defense posture, discussions are in full swing in Korea that AI policies should shift their focus from industrial development to national risk management.
The fuse is the new model ‘Mitos’ unveiled by Anthropic this month. According to the New York Times (NYT) on the 22nd (local time), Ansropic severely limited the scope of its disclosure because Mitos’ ability to find and attack hidden flaws in software that supports banks, power grids, and government systems around the world has reached a dangerous level. There is only one country other than the United States that has been allowed outside access.
The reason why the Mitos crisis is shocking is that a private company actually owns state-level cyber assets alone.
Anthropic shared Mitos with more than 40 organizations that operate global core infrastructure, revealing a list of 11 partner companies, including Amazon, Apple, and Microsoft (MS). All of these are American companies. The European Commission has met with Anthropic three times since its disclosure, but has yet to agree on a model approach. Claudia Platner, head of the German Federal Information Security Administration (BSI), also visited San Francisco without access to it and only listened to the explanation.
The more marginalized countries are, the greater the sense of crisis. The Chinese and Russian governments have not made an official statement, but many of China’s banks, energy and government agencies are running on the same software that Mitos has found vulnerabilities.
Matt Sheahan, a senior researcher at the Carnegie Endowment for International Peace, told the NYT, “The Mitos crisis will be China’s second wake-up call following the advent of ChatGPT.” Analysts say that the U.S. policy that has blocked exports of advanced semiconductors to China is widening the gap.
Despite this situation, there is virtually no international cooperation. This is because there is no international norm in the field of AI corresponding to the Nuclear Non-Proliferation Treaty (NPT), and there is no agreed verification procedure. This means that the right to access itself has become a diplomatic card.
Anthropic said it expects other companies to unveil models with similar cyber capabilities within 18 months. This also means that countries have been given a year and a half to overhaul their defense systems.
There’s already been a red flag. Anthropic said on the 21st that it was investigating after a report of unauthorized access to some versions of Mitos was received. As soon as the model itself leaks out, control disappears.
Even the U.S. government is not taking this issue lightly. “As the underlying AI model becomes more important, access rights themselves are moving into the realm of geopolitics,” Eduardo Levy Yeyatti, former chief economist at Argentina’s central bank, told the NYT.
The controversy surrounding Mitos is not just a global competition for technological supremacy, but is spreading to each country’s reality response stage. While the U.S. and Europe are checking the defense posture of financial and power grids, voices are calling for high-tech AI to be treated as a national security task rather than an industrial policy in Korea.
At the PwC Consulting Meeting held in Yeouido, Seoul on the 23rd, suggestions poured in that the domestic cyber defense system should be completely redesigned in the wake of the Mitos crisis.
Moon Hong-ki, CEO of PwC Consulting, said, “It is time to review the cybersecurity system from zero base.”
Lee Sang-geun, head of the AI Security Research Institute at Korea University, said, “Mitos is not a gradual improvement of the existing model, but a generational leap forward.” Director Lee pointed out, “There are concerns about information asymmetry as Korea is alienated from global cooperation systems such as project glaswing,” adding, “It is necessary to establish an independent response system.”
He said that emergency patches and the National Vulnerability Database (NVD) replacement system should be prepared as countermeasures. He then mentioned that it is time to urgently check the inventory of assets such as the country’s core infrastructure and financial sector. In order to defend itself, he stressed that security monitoring should be strengthened and global partnerships should be promoted.
[Silicon Valley = Correspondent Wonho-seop / Seoul = Reporter Park Sung-bae]
Click Here For The Original Source
