Are insecure code completions a vulnerability? — Seth Larson

Three months ago I saw that PyCharm shipped with a
“Full Line Completion” plugin that “uses a local deep
learning model to suggest entire lines of code”. These
suggestions manifest as whole-line suggestions after
you start typing and can be accepted with Tab. Essentially
auto-complete for entire lines.

I decide to test this functionality. I started by
writing import urllib3, created a new line,
and then typed u and received a suggested completion for the line
marked below with a
dashed border.
I was not impressed by the result:

import urllib3
urllib3.disable_warnings(urllib3.exceptions.InsecureRequestWarning)

Accepting this line would mean that any insecure
requests made with urllib3 would not result in a user-visible warning.
I didn’t accept this suggestion and then began to instantiate a
urllib3.PoolManager and what I feared would come next was confirmed:

import urllib3

urllib3.PoolManager(
    cert_reqs='CERT_NONE',

The suggestion offered to disable certificate verification (CERT_NONE) which
would make every request made by the PoolManager susceptible to
monster-in-the-middle (MITM) attacks. Accepting this code as-is would
mean the program I am writing has a severe vulnerability. If I
had accepted the prior suggestion too, then urllib3 would
have no chance to warn the user about this mistake prior
to productionizing this code.

Clearly something insecure is going on here, but for a CVE
to be assigned we have to decide which software component is
vulnerable. Does this behavior warrant a CVE at all? I am not sure
which is unfortunate, without a security-angle to a bug
report companies are less likely to prioritize reports.

I reported this behavior to JetBrains for “Full Line Code Completion” v253.29346.142
and clearly their support staff weren’t certain whether this defect
was a security vulnerability or not either. When I asked to
publish a blog post about this behavior after they confirmed
this report wasn’t a “direct security vulnerability” (which
I agree with) but then was asked not to publicize my report and referred to
PyCharm’s Coordinated Disclosure Policy
so… which is it? Security vulnerability or not?

I ended up waiting the 90 days anyway and I didn’t hear
back with any substantive update from the development team. I double-checked again
today using “Full Line Code Completion” v261.24374.152
and the behavior is identical, suggesting the same
insecure code for both contexts.

This isn’t meant to be a specific dig at PyCharm or JetBrains,
I have no-doubt that examples like this exist in every code generation
model available. I don’t think using CVEs for this purpose is
appropriate or helpful for users, either. But not prioritizing and addressing this
behavior at the source means more work to mitigate
the potential for insecure code to be accepted by users who are trusting
what is offered to them by their IDE.

What do you think? I am interested in knowing your thoughts
about this specific class of issue with code generation models.

Wow, you made it to the end!