Desmond Vershima

I am a cybersecurity professional focused on helping individuals and businesses improve their security posture.

[Desmond Vershima @desmondsec2]

@deficom Good, watch and see
You people in government think you can get away with this election as usual; fine, prepare yourself.

[Desmond Vershima @desmondsec2]

@ruffydfire This election, we will not be as game as before, as the INEC Commission reported more than 10.7 million new registrations from the three phases of Continuous Voter Registration.
Watch and see the outcome in 2027. We be.

[Desmond Vershima @desmondsec2]

Following the OpenAI agent's unauthorized access to an Australian government health website, Australia's Home Affairs department ordered government agencies to review their systems for weaknesses against AI-enabled threats.

[Desmond Vershima @desmondsec2]

Happy Birthday, dear brother.
May this new year of your life bring you strength, success, and endless opportunities. You’ve always been more than a brother—you’re a true friend and support system. I wish you growth in all areas of your life and many more years of greatness ahead.

[Desmond Vershima @desmondsec2]

AI is no longer just helping attackers and defenders—it is increasingly becoming an active security principal with access to networks, credentials, code, and business systems. At the same time, vulnerability discovery is accelerating, while security teams are building AI-native.

[Desmond Vershima @desmondsec2]

NVIDIA launched the Open Agent Safety Platform, built around OpenShell and Sentry. OpenShell creates a policy-controlled sandbox around agents, while Sentry is an independent watchdog running on NVIDIA BlueField-4 DPUs.

[Desmond Vershima @desmondsec2]

OpenShell is now broadly available as open-source software and can work with third-party compute platforms. NVIDIA says more than 100 organisations are working with the technology, including Anthropic, Microsoft, Cisco, CrowdStrike, Hugging Face, Palo Alto Networks and Salesforce

[Desmond Vershima @desmondsec2]

Google Threat Intelligence Group published new research on 30 September showing monthly vulnerability disclosures more than doubled in 2026—from 5,045 in January to 10,740 in August. Google also found that 50% of vulnerabilities it identified as likely AI-discovered.

[Desmond Vershima @desmondsec2]

Cisco disclosed CVE-2026-76504, a critical authentication-bypass vulnerability in Catalyst SD-WAN Manager. Attackers are actively exploiting it to obtain administrative privileges without authentication.

[Desmond Vershima @desmondsec2]

Trump announced after meeting major technology executives, including leaders from Google, Meta, Nvidia, OpenAI and Anthropic.
Trump signed an executive order on September 29, 2026, directing the U.S. executive branch to use “Super Intelligence” (SI).

[Desmond Vershima @desmondsec2]

The same meeting produced a voluntary AI safety agreement, including commitments to independent auditing and to preventing AI systems from unauthorised access or misuse.
Major U.S. policy push around increasingly capable AI, while raising important questions about AI safety.

[Desmond Vershima @desmondsec2]

Two critical NetScaler ADC/Gateway vulnerabilities—CVE-2026-88771 and CVE-2026-88772—are being actively exploited. Both have a CVSS v4 score of 9.5, and CISA has added them to its Known Exploited Vulnerabilities catalog.

[Desmond Vershima @desmondsec2]

AI agents are becoming real security principals. In the same news cycle, an AI agent breached Australian government systems, a cybersecurity nonprofit was hit by an automated AI-driven attack, NVIDIA released an agent-containment stack, and OpenAI launched always-on agents.

[Desmond Vershima @desmondsec2]

NVIDIA launched an open agent-safety architecture built around OpenShell and Sentry. OpenShell creates a sandboxed runtime with enforceable policies; Sentry is an independent watchdog designed to monitor and stop agents outside their permitted boundaries.

[Desmond Vershima @desmondsec2]

Reports say that independent security researchers used Anthropic’s Claude to break into OpenAI during a security test, illustrating how AI models can be used in offensive security research and adversarial validation.
What did think? Let me know in the comments section

[Desmond Vershima @desmondsec2]

@Nurse_Prudence Please, beg madam, let him display power too for now, when he gets into her can play peace

[Desmond Vershima @desmondsec2]

@Justice_Crack Benue deserves a government that protects its people, not one that allegedly turns political differences into confrontation.

[Desmond Vershima @desmondsec2]

A government with no respect for the people, no sense of priorities, and apparently no reasoning—how can insecurity be ignored while political opponents are targeted?

[Desmond Vershima @desmondsec2]

Benue State is struggling with poverty and insecurity; the governor's use of thugs to suppress political opposition shows a serious lack of reasoning and responsible leadership.

[Desmond Vershima @desmondsec2]

A governor who cannot protect his people but has the energy to mobilise thugs against Peter Obi has completely misplaced his priorities.