https://community.checkpoint.com/t5/AI-Agents-Security/Lakera-News-A-New-Attack-Path-That-Surprised-Us-This-year/m-p/266664#M44 <P>Excellent!</P> Wed, 07 Jan 2026 11:51:00 GMT the_rock 2026-01-07T11:51:00Z https://community.checkpoint.com/t5/AI-Agents-Security/Lakera-News-A-New-Attack-Path-That-Surprised-Us-This-year/m-p/266663#M43 <P>Hey CheckMates!<BR /><SPAN>One of the themes that kept coming up in the last few weeks is how quickly <STRONG>indirect attacks</STRONG> are growing.</SPAN></P> <P><SPAN>These look harmless at first:</SPAN></P> <UL> <LI><SPAN><EM>“Can you summarize this link?”</EM></SPAN></LI> <LI><SPAN><EM>“Can you extract key points from this document?”</EM></SPAN></LI> </UL> <P><SPAN>But the malicious instruction is hidden inside the source. Once the model reads it, it becomes part of its context.</SPAN></P> <P><SPAN>In our data, these attacks required <STRONG>fewer attempts</STRONG> than direct prompt injections. That was one of the clearest signals we saw this year.</SPAN></P> <P><SPAN>If you want the deeper dive, <A href="https://d31-0l04.eu1.hubspotlinks.com/Ctc/L0+113/d31-0L04/VVSw-l82TrJbW7tzSVX8P6thSW6TYKFS5H_g5JN3hBC585nXHCW50kH_H6lZ3pPVt99pw6yZdmyW2_p5vR8LGD7FW8453gt8PG8qhW5jzCz66kZLzwW2_THSW1Jt2TqW98tj8s6q5FQbW5SnvcF5KZWZYW6PCxbg7TgpV3N7Z7_CCgpPVTW9bGrgd1KRwHnW4j4JwM1symxtW2yp6-T35p1B1W4mq_S625vT96V-s8W33f7vbtW1nwT-59hW6zNW98_-dL4qZLR6W2m6QcC5RdbRPW53f2jW84bbkhW5bTQ6l1zscTqW6mlbqk6bCLzHW2tPJtp2TWXjXW1s5MjJ3rG2_LW9klL0X4VjwWGW4ltB8Z2FCvVMW10ZrB88R9ZMfW57vmhn8lhMX-VJV4r-4fYC_1N5TzW5s5ky2rW6SyVx58s6cMwVXxlSq5PmT9fW9kmF5y4Q_rBVW76MVdj3k_zGff990Rr604" target="_blank" rel="noopener" data-hs-link-id="0" data-hs-link-id-v2="FFP/rj8n">page 3 of Lakera's Q4 2025 Agent Security Trends Report covers it</A>.</SPAN></P> <P><SPAN>If you are experimenting with browsing, RAG, agent tools, or connected workflows, we can walk you through how we secure these paths in real deployments. It is usually a 15-minute conversation.</SPAN></P> <H2 class="lia-align-center"><A href="https://www.lakera.ai/contact?utm_campaign=240643310-25-Q4%20Year%20of%20the%20Agent%20Webinar" target="_self">Talk to our AI Security Team</A></H2> Wed, 07 Jan 2026 11:45:54 GMT https://community.checkpoint.com/t5/AI-Agents-Security/Lakera-News-A-New-Attack-Path-That-Surprised-Us-This-year/m-p/266663#M43 _Val_ 2026-01-07T11:45:54Z https://community.checkpoint.com/t5/AI-Agents-Security/Lakera-News-A-New-Attack-Path-That-Surprised-Us-This-year/m-p/266664#M44 <P>Excellent!</P> Wed, 07 Jan 2026 11:51:00 GMT https://community.checkpoint.com/t5/AI-Agents-Security/Lakera-News-A-New-Attack-Path-That-Surprised-Us-This-year/m-p/266664#M44 the_rock 2026-01-07T11:51:00Z