As reported by SecurityAffairs in their Malware Newsletter Round 115, this week's threat landscape paints a sobering picture of converging attack vectors that demand defender attention. Several distinct threads emerge from the collection of research highlights — and each carries implications that extend well beyond the individual incidents.

Threat Alert: As reported by SecurityAffairs in their Malware Newsletter Round 115, this week's threat landscape paints a sobering picture of converging attack vectors that demand defender attention.

Three Trends Defenders Cannot Ignore

1. AI is now both weapon and target. The newsletter references multiple AI-related threats: RatHat, an AI-powered mobile threat targeting credentials and bank accounts; skill poisoning attacks that turn AI agents into malware droppers (flagged by China's own National CERT); and academic research on adversarial attacks against LLM-based malware analyzers (ALIBI). This is no longer theoretical. Threat actors are actively building AI into their tooling while simultaneously probing AI security systems for weaknesses. Security teams that have deployed AI-powered detection or autonomous response tools should assume these systems are themselves under attack.

2. Supply chain attacks have reached industrial scale. The Brevo supply chain attack reportedly compromised over 100,000 sites with WordPress backdoors and Clickfix malware. This follows a now-familiar pattern: compromise a trusted upstream provider, inject malicious payloads downstream, and let the trust relationship do the distribution work. The scale here is what matters — 100,000 sites is not a targeted campaign, it's a mass-infection event that creates persistent infrastructure for future operations.

3. Browser extensions and OAuth tokens remain an underdefended perimeter. The malicious Twitch extension exposing 30,000 users' OAuth tokens to a Russian bot service, combined with the KREMLIN campaign forging Chrome integrity checks to steal banking sessions, illustrates that the browser — and especially its extension ecosystem — remains one of the most exploitable surfaces in enterprise environments. OAuth tokens grant persistent access without requiring password re-entry, making them high-value targets that MFA does not protect.

What the Geography Tells Us

The newsletter's geographic spread is notable. Iranian threat actors are targeting dissidents and journalists. Chinese groups are chaining Chrome and Windows zero-days and operating MQTT-based infection brokers across Asia. Russian-linked infrastructure is harvesting OAuth tokens. Central Asian energy and government sectors face SilkParasite/SpiceRAT campaigns. This is not random noise — it reflects a world where geopolitical tensions increasingly manifest through cyber operations against both hardened targets and soft civilian infrastructure.

The common thread across these campaigns is exploitation of trust: trust in software supply chains, trust in browser extensions, trust in AI systems, and trust in geopolitical quiet. Defenders should audit where their trust assumptions are most exposed.

Shield53 Recommendations

Shield53 Recommendations
Audit AI tooling exposure: Inventory every AI-powered security or productivity tool in your stack. Assess whether it could be poisoned, bypassed, or leveraged as a dropper. Apply least-privilege to AI agents and monitor their API calls for anomalous patterns.
Tighten extension governance: Enforce allowlists for browser extensions across managed devices. Revoke and rotate OAuth tokens for any service that has had a public extension compromise. Consider blocking OAuth token reuse from unmanaged devices entirely.
Harden supply chain monitoring: For WordPress and CMS-dependent organizations, implement file integrity monitoring on plugin directories and subscribe to upstream vendor security notifications. The Brevo incident demonstrates that compromise can persist for weeks before detection.
Patch n-day exposure: The Red Heron campaign exploiting a Gitea n-day flaw with a new Linux rootkit is a reminder that known vulnerabilities remain highly effective. Prioritize patching of development infrastructure — Git servers, CI/CD systems, and container registries are high-value targets that often lag in patching cycles.
Monitor MQTT egress: If your organization allows MQTT traffic, monitor for unauthorized broker connections. Threat actors increasingly use MQTT as a lightweight C2 channel because it blends in with IoT and messaging infrastructure.

The overarching lesson from this week's landscape: attackers are scaling through trust, automating through AI, and diversifying through geography. Defenders must match that breadth — not by chasing every headline, but by systematically reducing the trust surfaces that make these campaigns possible.