Page 10 - The Indian EYE 052926

OPINION MAY 29, 2026 | The Indian Eye 10

The Making of a Mythos:

AI-Enabled Cybersecurity and the

Emerging Architecture of Access

Anthropic’s release of the frontier AI model Claude Mythos Preview has caused significant trepidation, as its autonomous

code-reasoning capabilities surfaced previously undetected vulnerabilities across major operating systems and software
infrastructure. India needs to secure critical infrastructure by updating, modernising or replacing legacy systems

MEGHNA PRADHAN

n April 2026, Anthropic stated
that its frontier artificial intel-
Iligence model, dubbed Claude
Mythos Preview, had unprecedented
cyber capabilities that made it too
dangerous to release publicly. They
disclosed that the Large Language
Model (LLM), during its testing
phase, had identified ‘thousands of
high and critical severity vulnerabili-
ties’, many of which have been, and
continue to be, validated by cyberse-
curity experts. These include discov-
ering and exploiting a 27-year-old
Zero Day Vulnerability (0-day) in
OpenBSD, finding a 16-year-old bug
in media processing library FFMpeg,
and patching 271 vulnerabilities in
Firefox 150 based on a single evalu-
ation cycle.
While Mythos Preview’s ca-
pabilities mark a significant step
in the development of AI systems,
their potential has rung alarm bells For India, Mythos is not an issue because it creates a novel vulnerability, but because it significantly compresses
among global cybersecurity com- the timeline for existing threats (File photo)
munities, financial institutions and
governments. The global financial launched Project Glasswing, an AI es important concerns. Access to a US had initially restricted Anthropic
elites have especially highlighted consortium comprising companies system with such glaring implications from expanding the number of en-
cyber threats posed by Mythos. An- such as Amazon Web Services, Ap- for the security of global financial and tities for strategic reasons. The US
thropic has disclosed that Mythos ple, Broadcom, Cisco, CrowdStrike, cyber systems is being administered Commerce Department’s Center for
was used to conduct what could es- Google, JPMorgan Chase, the Linux at the discretion of a private entity AI Standards and Innovation, along
sentially be called ‘digital robbery’ Foundation, Microsoft, NVIDIA, to other private entities, without a with Anthropic, OpenAI, and Goo-
on systems across the world, which and Palo Alto Networks, as well as statutory framework, treaty obliga- gle, is also discussing the potential
has led to fears that the system may over 40 other organisations with ac- tions, or accountability mechanisms. development of a pre-deployment
further be misused for mass looting cess to the system. Additionally, the selected partners evaluation framework.
of bank accounts, paralysis in inter- Anthropic’s response to Mythos’s are, barring a few exceptions, largely The Mythos reveal has sent
national payment systems, or spark capabilities has been to create an AI American companies, most of which alarm bells ringing across the globe,
a crisis of confidence in extant fi- consortium consisting of some of the already have access to advanced AI and India is no exception. On 23
nancial systems. major big tech companies and finan- capabilities and strong cybersecurity. April 2026, Finance Minister Nirma-
Global finance leaders and cen- cial institutions. On the surface, the There are additional reports that la Sitharaman convened a high-lev-
tral bankers from countries such as logic of Project Glasswing remains Anthropic has been engaging with el meeting. Minister of Electronics
the US, Canada, the EU, the UK, defensible: a public release of the US national-security institutions to and Information Technology Ashwi-
Germany, South Korea and India LLM would give malicious actors negotiate terms of access to Mythos, ni Vaishnaw, heads of banking and
have expressed concerns and inter- the tools to damage structures that possibly as part of a broader effort to key cybersecurity bodies assessed
est in accessing the LLM to plug the current cybersecurity apparatus rebuild relationships strained since Mythos-related cybersecurity risks to
vulnerabilities in their financial sys- is not equipped to defend. However, the fallout from the earlier contro- financial institutions. While deeming
tems. Simultaneously, Anthropic has the structure of this project also rais- versial defence AI partnership. The Continued on next page... >>

www.TheIndianEYE.com