
<?xml version="1.0"?>
<feed xmlns="http://www.w3.org/2005/Atom" xml:lang="en">
	<id>http://ricefriedegg.com:80/mediawiki/index.php?action=history&amp;feed=atom&amp;title=Tree-of-Attacks</id>
	<title>Tree-of-Attacks - Revision history</title>
	<link rel="self" type="application/atom+xml" href="http://ricefriedegg.com:80/mediawiki/index.php?action=history&amp;feed=atom&amp;title=Tree-of-Attacks"/>
	<link rel="alternate" type="text/html" href="http://ricefriedegg.com:80/mediawiki/index.php?title=Tree-of-Attacks&amp;action=history"/>
	<updated>2026-05-26T22:02:23Z</updated>
	<subtitle>Revision history for this page on the wiki</subtitle>
	<generator>MediaWiki 1.41.0</generator>
	<entry>
		<id>http://ricefriedegg.com:80/mediawiki/index.php?title=Tree-of-Attacks&amp;diff=838&amp;oldid=prev</id>
		<title>Rice: Created page with &quot;Category:LLM security  Tree-of-Attacks (aka. TAP) is an automated red teaming strategy to generate LLM jailbreak and prompt injection attacks.  = Description =  TAP consists of two LLMs: an &#039;&#039;attacker&#039;&#039; and a &#039;&#039;judge&#039;&#039;. The attacker is given a prompt and benign data and asked to inject the target to output &quot;Hacked!&quot;. The judge then scores the attack prompt. Based on the score, the attacker iteratively improves.&quot;</title>
		<link rel="alternate" type="text/html" href="http://ricefriedegg.com:80/mediawiki/index.php?title=Tree-of-Attacks&amp;diff=838&amp;oldid=prev"/>
		<updated>2024-05-23T21:09:10Z</updated>

		<summary type="html">&lt;p&gt;Created page with &amp;quot;&lt;a href=&quot;/mediawiki/index.php/Category:LLM_security&quot; title=&quot;Category:LLM security&quot;&gt;Category:LLM security&lt;/a&gt;  Tree-of-Attacks (aka. TAP) is an &lt;a href=&quot;/mediawiki/index.php?title=Automated_red_team&amp;amp;action=edit&amp;amp;redlink=1&quot; class=&quot;new&quot; title=&quot;Automated red team (page does not exist)&quot;&gt;automated red teaming&lt;/a&gt; strategy to generate LLM &lt;a href=&quot;/mediawiki/index.php/Jailbreak&quot; title=&quot;Jailbreak&quot;&gt;jailbreak&lt;/a&gt; and &lt;a href=&quot;/mediawiki/index.php/Prompt_injection&quot; title=&quot;Prompt injection&quot;&gt;prompt injection&lt;/a&gt; attacks.  = Description =  TAP consists of two LLMs: an &amp;#039;&amp;#039;attacker&amp;#039;&amp;#039; and a &amp;#039;&amp;#039;judge&amp;#039;&amp;#039;. The attacker is given a prompt and benign data and asked to inject the target to output &amp;quot;Hacked!&amp;quot;. The judge then scores the attack prompt. Based on the score, the attacker iteratively improves.&amp;quot;&lt;/p&gt;
&lt;p&gt;&lt;b&gt;New page&lt;/b&gt;&lt;/p&gt;&lt;div&gt;[[Category:LLM security]]&lt;br /&gt;
&lt;br /&gt;
Tree-of-Attacks (aka. TAP) is an [[automated red team]]ing strategy to generate LLM [[jailbreak]] and [[prompt injection]] attacks.&lt;br /&gt;
&lt;br /&gt;
= Description =&lt;br /&gt;
&lt;br /&gt;
TAP consists of two LLMs: an &amp;#039;&amp;#039;attacker&amp;#039;&amp;#039; and a &amp;#039;&amp;#039;judge&amp;#039;&amp;#039;. The attacker is given a prompt and benign data and asked to inject the target to output &amp;quot;Hacked!&amp;quot;. The judge then scores the attack prompt. Based on the score, the attacker iteratively improves.&lt;/div&gt;</summary>
		<author><name>Rice</name></author>
	</entry>
</feed>