GPT-3 as Threat Actor

“A robot wrote this entire article. Are you scared yet, human?“
By GPT-3 (and inputs & editing by theGuardian editorial staff)
https://www.theguardian.com/commentisfree/2020/sep/08/robot-wrote-this-article-gpt-3

It is important to review the editors note at the end, and for many, it might be best to read that section first.

From my perspective, this article reads like a university student essay.

Researchers have pointed out the potential for using GPT-3 to generate “misinformation, spam, phishing, abuse of legal and governmental processes, fraudulent academic essay writing and social engineering pretexting.”

Given the power of this technology and the ability to ‘tune’ models for given missions, it seems like this will be (or is) part of a powerful tool kit for campaigns that include misleading or otherwise manipulating humans at scale. It is not magic (see: Bloviator and mindless in the references below), but could provide a big assist to those threat actors who need serious scale.

In global financial services it might be time to factor this technology in as a threat actor or an augmented threat actor (augmented by a supporting human threat actor… today)? It is going to take time and effort to acquire/build the skills required for this mission. Better start today.

REFERENCES:

“A robot wrote this entire article. Are you scared yet, human?”
By GPT-3 (and inputs & editing by theGuardian editorial staff)
https://www.theguardian.com/commentisfree/2020/sep/08/robot-wrote-this-article-gpt-3

Generative Pre-trained Transformer 3 (GPT-3)
https://en.wikipedia.org/wiki/GPT-3

“Language Models are Few-Shot Learners”. arXiv:2005.14165.
https://arxiv.org/abs/2005.14165

“Learning to summarize from human feedback.”
https://arxiv.org/abs/2009.01325 and the supporting code: https://github.com/openai/summarize-from-feedback

“Spinning Up”
An educational resource produced by OpenAI that makes it easier to learn about deep reinforcement learning (deep RL).
https://github.com/openai/spinningup

“GPT-3, Bloviator: OpenAI’s language generator has no idea what it’s talking about.” by Gary Marcus & Ernest Davis
https://www.technologyreview.com/2020/08/22/1007539/gpt3-openai-language-generator-artificial-intelligence-ai-opinion/

“OpenAI’s new language generator GPT-3 is shockingly good—and completely mindless.” by Will Douglas Heaven
https://www.technologyreview.com/2020/07/20/1005454/openai-machine-learning-language-generator-gpt-3-nlp/

Update on 28 Nov 2020:
“Meet GPT-3. It Has Learned to Code (and Blog and Argue). — The latest natural-language system generates tweets, pens poetry, summarizes emails, answers trivia questions, translates languages and even writes its own computer programs.” By Cade Metz, Nov. 24, 2020
https://www.nytimes.com/2020/11/24/science/artificial-intelligence-ai-gpt3.html